[dwk]

Yes, it's an endless debate, but apparently adc doesn't realize that. Acd - you are not the first to wade in to the audio world claiming specs and science have 'solved' the audio reproduction problem. People were probably doing that back in the 60's. Whether you intend it this way or not, the basic upshot of your position is this:

"I know everything there is to know about the human auditory system"

Sounds like a statement that might be overreaching *just* a bit. Now, I consider myself largely on the 'objective' side of the O/S debate, and the majority of my decisions are technically guided. I probably believe that about 95+% of the anecdotal reports of 'this is better' are purely psychological in nature and thus don't pay any attention to them. BUT, it still doesn't take much digging to realize that there is far more at play in how we hear and percieve music than we currently understand. It also doesn't take much digging to realize that the entire idea of formal testing auditory phenemonena is fraught with practical difficulties, however simple it may look on paper.

So, given that so many folks have seen this type  of thing go by before, when you wade in with overly simple absolute statements on audibility it says to most people that you haven't really done your homework.

Ultimately, there are limits to what we can prove about audibility. Those limits currently leave open a lot of questions, and whenever that happens you end up dealing with a situation based on faith or belief, neither of which are provable pretty much by definition (though they obviously are arguable in the sense that we still argue about them). So, it's pointless.  The best thing that (hopefully) can come out of this is that by approaching certain aspects with both skepticism and an open mind, we can theorize about some characteristics that ARE isolatable to the point of being able to start testing for them; this will undoubtedly be a slow process.

To address one specific example:

You guys aren't actually reading what I've written. An engineer/designer can alter the circuit topography of his amp to make it sound like a tube amp. If he can do that, it means there's no magic to tube design. It also means that the data can describe sound.

Sounds great, except that it's wrong in that it presupposes it's own conclusion.  Tube amps differ from solid state amps in MANY ways, not just the shape of the transfer function. They have different output impedance characteristics. They have different distortion levels AS WELL AS spectral distribution. They have different saturation/overload characteristics. The output xformer introduces different bandwidth constraints. They'll almost certainly have different EMI/RFI signatures and sensitivities. Furthermore, all of this may change with frequency and/or absolute reproduction level.
 So, exactly which of those is responsible for 'tube sound'?  We don't know, and it's impossible for an SS amp to simultaneously match a tube amp in all of these aspects. Certainly things like the Zen amps seem to have similar characteristics to simple tube topologies, but even then there are differences.

So, no - there is no 'magic'. There isn't any 'ether' through which the 'soul' of the music travels independent of the electrical circuits. But the truth is that it's a heck of a lot harder to identify which characteristics of that signal correspond to 'music' than it first appears, and thus it's fiendishly difficult to account for what listeners report that they hear.

Having said all that, I have to wonder whether this is at least partially a troll. Someone that sells artistic goods on their web site (which at a quick glance actually look pretty cool) and owns an RWA amp doesn't strike me as the typical profile for someone advancing the 'numbers are everything' argument.

[miklorsmith]

Nice post!   

[darrenyeats]

If I may put the other side of the argument... (You didn't think this thread was going to end there did you!?)

Existing measurement techniques support the fact that SS amps are more accurate than valves. If anyone has information to the contrary I would love to read it.

If you are saying valves sound better to you I am just fine with that. If you are saying valves are more accurate but in an unknown way, maybe you can win the Nobel prize by figuring out what that way is. Until then you're not able to rely on, or even hide behind, any figures.

I already said if you like valves then fine. The issue then becomes, if someone with golden ears or a manufacturer says valve amps are better in some sort of objective way (not measurable) should I believe him/her? This is where I would say prove it with some kind of blind test. Without a blind test I will file claims of superiority under "live and let live". With a blind test, I will file claims of superiority under "has been shown - and measurements be damned!".

Blind testing is a pain in the butt but it's all we've got for nailing the placebo effect. Again, if you figure out an easier way to let people listen without bias, you could get yourself a Nobel prize. As I said the problem is only a thorny one for people like reviewers or manufacturers since what anyone prefers in the privacy of their own home is nobody else's business. But we are discussing more than that when we talk about the Benchmark vs modders etc.
Darren

[miklorsmith]

There should be one thread or one circle where blind testing arguments are allowed and nowhere else.  It's the dead-endedest topic in all the forums for all time and has no place in any other topic except "Let's talk blind testing".

[darrenyeats]

Well, I guess I could just go away and haunt hydrogenaudio instead.

[miklorsmith]

Just my opinion of course, but this is the classic immovable object versus the irresistable force - a guaranteed entrenchment with no movement on either side.

[darrenyeats]

Once I heard a nice quote. "By the age of 40 you've managed to surround yourself with people who agree with you."

It's good that we can all coexist and challenge each other - even if we don't change our minds. It makes life more interesting.
Darren

[dwk]

If I may put the other side of the argument... (You didn't think this thread was going to end there did you!?)

Existing measurement techniques support the fact that SS amps are more accurate than valves. If anyone has information to the contrary I would love to read it.

If you are saying valves sound better to you I am just fine with that. If you are saying valves are more accurate but in an unknown way, maybe you can win the Nobel prize by figuring out what that way is. Until then you're not able to rely on, or even hide behind, any figures.

Darren

I don't have time for a full reply right now, but I'll throw some thoughts out there. I certainly don't have any 'answers' - just speculation and ideas.

I'm at the point that I think we need to consider what 'accuracy' means. Audio reproduction is not a 1-dimensional problem, and the perception of audio/music is not only multi-dimensional but the dimensions are not even obvious.  Even taking a 'simple' metric like THD interpreting it is tough - audibility of harmonic distortion depends on a lot of factors - order of the distortion, frequency of the fundamental, absolute playback level, AND the musical content of the signal due to masking phenomena.

So, what does it mean to be 'accurate'? It's tempting to put aside the perceptual element since it's so intractable, but unfortunately it's really the only important part. The perceptual coding guys have pretty much proven that - 320k VBR compresion throws away 75+% of the raw signal  (and even 50+% of the entropy measured information) and yet still retains something pretty darn close to audibly transparent. It's absolutely horrible from any simple technical measure, but yet very 'accurate'.

So, my feeling is that we need to start thinking in 'perceptual space' rather than 'technical space'. AND we have to stop thinking 'steady state' and start thinking 'musically'.

In a nutshell, what I think we need to do is 'invert' the lossy perceptual coding idea to try to measure system deviations against these perceptual dimensions. This is inherently a time-varying problem though, so we end up with a multi-dimensional time-varying error signal which we somehow have to integrate into a 'score'. Not easy

[audio-heaven]

Against my better judgement I'll chime in with a few thoughts I own a Benchmark Dac1 and although I think it is a very decent DAC for the money it is by no means the be all and end of all DACs despite how well it measures from a technical standpoint.

I have recently built myself a RAKK DAC MkII that sounds FAR superior to my ears than the Benchmark DAC1 in every department! Does this mean it measures WAY better than the Benchmark DAC1?

There are certainly some quite large differences between the 2 DACs for one example the Benchmark has op amps on the output stage where the RAKK DAC has a pair of output transformers. How much difference would some op transformers make to the Benchmarks sound? Some people think that output transformers sound better and are more transparent sounding than any op amp around, so does this mean the Benchmark could be improved by using OP transformers instead of op amps?

The huge problem with getting a handle on any single components 'sound' are multifarious at best not least the astronomical combinations in partnering equipment all with different impedances not to mention the acoustic environments they are experienced in.
This is why you can read often very conflicting reviews from reviewers (professional or public) on the same piece of gear. How much of their opinion is simply down to personal preference and how much are all the 'other' factors forming the verdict.

Ignorance is bliss to a large extent, and a lot depends on personal experiences, if someone has heard a relatively small amount of decent gear and the Benchmark DAC1 was the best thing they had heard to date than their review of it would probably be very favourable.
The fact that all systems at any price have differences in sound proves (to me at least) we are still a very long way from making the PERFECT sounding system through technical measurements alone. Even if were possible taking into account that everyone has different listening rooms (some kind of room correction might be in order here ) would EVERYBODY love the result?

AH

[darrenyeats]

So, what does it mean to be 'accurate'? It's tempting to put aside the perceptual element since it's so intractable, but unfortunately it's really the only important part. The perceptual coding guys have pretty much proven that - 320k VBR compresion throws away 75+% of the raw signal  (and even 50+% of the entropy measured information) and yet still retains something pretty darn close to audibly transparent. It's absolutely horrible from any simple technical measure, but yet very 'accurate'.

AH, that's an interesting argument. Are you saying valves are good in these unknown perceptual dimensions? Regardless of these unknowns, there is still some way to go before valve amps do well with existing measurements which are known to correspond to audible sound quality. However, I admit valve amps have seemed to get better (based on existing measurements) as time has gone by.

An honest question now. What if someone designed a valve amp that measured as being really accurate according to _existing_ measurements. Do you find that idea appealing?
Darren

[miklorsmith]

Sure.  Not exactly the same, but John Atkinson measured my Lamm preamp and found it exemplary.  Of course, I paid more attention to the listening notes.   

Dan Wright makes some tubed preamps that measure great.  I'm sure there are others.

Of course, JA also measured my Yamamoto SET amp and practically thought it was broken.  I love the sound of it, like just about everyone else that has on the right speakers.

[doug s.]

a couple questions - what does this have to do w/the benchmark dac sounding as good as it gets, cuz its specs are so good?   

and, while i agree that s/s gear in general has less distortion than tubed gear, there's *lotsa* tubed gear w/distortion <1%, sometimes way under 1%.  with everything else going on in an audio circuit, this is plenty low enough so as to be inaudible on its face - ie: it won't be a contributor to poor sound.  so, a tubed amp, with, say 0.5% distortion at 20w output may sound a lot better than a solid state amp at 20w output, even tho the s/s amp has orders of magnitude less distortion.  because the tube amp, w/its 0.5% distortion is still so low that it contributes no negative impact, due solely to distortion...

ymmv,

doug s.

If I may put the other side of the argument... (You didn't think this thread was going to end there did you!?)

Existing measurement techniques support the fact that SS amps are more accurate than valves. If anyone has information to the contrary I would love to read it.

If you are saying valves sound better to you I am just fine with that. If you are saying valves are more accurate but in an unknown way, maybe you can win the Nobel prize by figuring out what that way is. Until then you're not able to rely on, or even hide behind, any figures.

I already said if you like valves then fine. The issue then becomes, if someone with golden ears or a manufacturer says valve amps are better in some sort of objective way (not measurable) should I believe him/her? This is where I would say prove it with some kind of blind test. Without a blind test I will file claims of superiority under "live and let live". With a blind test, I will file claims of superiority under "has been shown - and measurements be damned!".

Blind testing is a pain in the butt but it's all we've got for nailing the placebo effect. Again, if you figure out an easier way to let people listen without bias, you could get yourself a Nobel prize. As I said the problem is only a thorny one for people like reviewers or manufacturers since what anyone prefers in the privacy of their own home is nobody else's business. But we are discussing more than that when we talk about the Benchmark vs modders etc.
Darren

[doug s.]

according to published specs, the distortion on my tubed preamp is 0.15% - low enuff for me...

speakers have distortion so many orders of magnitude higher than even tubed gear, it seems any decently designed component - tubed or solid state, shouldn't have distortion as its stumbling block...

doug s.

Sure.  Not exactly the same, but John Atkinson measured my Lamm preamp and found it exemplary.  Of course, I paid more attention to the listening notes.   

Dan Wright makes some tubed preamps that measure great.  I'm sure there are others.

Of course, JA also measured my Yamamoto SET amp and practically thought it was broken.  I love the sound of it, like just about everyone else that has on the right speakers.

[acd483]

Yes, it's an endless debate, but apparently adc doesn't realize that. Acd - you are not the first to wade in to the audio world claiming specs and science have 'solved' the audio reproduction problem.

I didn't claim that at all. I haven't heard a truly convincing reproduction of music as of yet. Do I think a DAC can be engineered to audible transparency...yes, do I think that solves the challenge of audio reproduction? No. Hardly infact!

People were probably doing that back in the 60's. Whether you intend it this way or not, the basic upshot of your position is this:

"I know everything there is to know about the human auditory system"

Sounds like a statement that might be overreaching *just* a bit.

Sorry you've inferred that. I most certainly do not. I think science has proven the audible range of the human ear. Again, no magic there.

To address one specific example:

You guys aren't actually reading what I've written. An engineer/designer can alter the circuit topography of his amp to make it sound like a tube amp. If he can do that, it means there's no magic to tube design. It also means that the data can describe sound.

Sounds great, except that it's wrong in that it presupposes it's own conclusion.  Tube amps differ from solid state amps in MANY ways, not just the shape of the transfer function. They have different output impedance characteristics. They have different distortion levels AS WELL AS spectral distribution. They have different saturation/overload characteristics. The output xformer introduces different bandwidth constraints. They'll almost certainly have different EMI/RFI signatures and sensitivities. Furthermore, all of this may change with frequency and/or absolute reproduction level.
 So, exactly which of those is responsible for 'tube sound'?  We don't know, and it's impossible for an SS amp to simultaneously match a tube amp in all of these aspects. Certainly things like the Zen amps seem to have similar characteristics to simple tube topologies, but even then there are differences.

I'm going to defer to Carver on this who claims that 90% of tube sound comes down to output impedance. He also refers to those in the field that have pretty easily replicated the sound of tube amp with transistors.

So, no - there is no 'magic'. There isn't any 'ether' through which the 'soul' of the music travels independent of the electrical circuits. But the truth is that it's a heck of a lot harder to identify which characteristics of that signal correspond to 'music' than it first appears, and thus it's fiendishly difficult to account for what listeners report that they hear.

True enough!

Having said all that, I have to wonder whether this is at least partially a troll. Someone that sells artistic goods on their web site (which at a quick glance actually look pretty cool) and owns an RWA amp doesn't strike me as the typical profile for someone advancing the 'numbers are everything' argument.

Partially a troll...that's funny. No I'm fully an artist though and I appreciate the kind words on my work. If anyone needs design work! Honestly though, I think there's enough soul in music that I don't need gear that warms it up or softens its edges. Think about guitarists who own hundreds and pick just the right guitar for a certain song...I'm not going to willingly alter that intention. That's the artist in me I guess.

So nobody liked my idea for the engineer test?

[acd483]

Once I heard a nice quote. "By the age of 40 you've managed to surround yourself with people who agree with you."

It's good that we can all coexist and challenge each other - even if we don't change our minds. It makes life more interesting.
Darren

As Dennis Miller quips "I like the cut of your jib man"

[acd483]

Slight tangent...if anyone is into Alison Krauss, her new album with Robert Plant, "Raising Sand", is awesome. I'm listening to this cut "Let Your Loss Be Your Lesson" (slightly apropos eh?) and it is groovin'.

[audioengr]

If I may put the other side of the argument... (You didn't think this thread was going to end there did you!?)

Existing measurement techniques support the fact that SS amps are more accurate than valves. If anyone has information to the contrary I would love to read it.

If you are saying valves sound better to you I am just fine with that. If you are saying valves are more accurate but in an unknown way, maybe you can win the Nobel prize by figuring out what that way is. Until then you're not able to rely on, or even hide behind, any figures.

Darren

I don't have time for a full reply right now, but I'll throw some thoughts out there. I certainly don't have any 'answers' - just speculation and ideas.

I'm at the point that I think we need to consider what 'accuracy' means. Audio reproduction is not a 1-dimensional problem, and the perception of audio/music is not only multi-dimensional but the dimensions are not even obvious.  Even taking a 'simple' metric like THD interpreting it is tough - audibility of harmonic distortion depends on a lot of factors - order of the distortion, frequency of the fundamental, absolute playback level, AND the musical content of the signal due to masking phenomena.

So, what does it mean to be 'accurate'? It's tempting to put aside the perceptual element since it's so intractable, but unfortunately it's really the only important part. The perceptual coding guys have pretty much proven that - 320k VBR compresion throws away 75+% of the raw signal  (and even 50+% of the entropy measured information) and yet still retains something pretty darn close to audibly transparent. It's absolutely horrible from any simple technical measure, but yet very 'accurate'.

So, my feeling is that we need to start thinking in 'perceptual space' rather than 'technical space'. AND we have to stop thinking 'steady state' and start thinking 'musically'.

In a nutshell, what I think we need to do is 'invert' the lossy perceptual coding idea to try to measure system deviations against these perceptual dimensions. This is inherently a time-varying problem though, so we end up with a multi-dimensional time-varying error signal which we somehow have to integrate into a 'score'. Not easy

dwk - you've hit the nail on the head IMO.  The so called "classical" measurement techniques are insufficient with todays low-noise high-bandwidth systems IMO.  A much more dynamic and stereo stimulus is required and then time-domain measurements as well as spectral: amplitude, frequency and phase.  And I believe the measurements need to take into account frequencies well above audibility.  Three dimensional plots may even be necessary to characterize systems properly.

I read with great interest every review that Stereophile does and all of the measurements that JA does.  Time and time again, I listen to these devices and the limited measurements just dont characterize these components sufficiently to tell whether they are "musical" or whether the imaging is any good - depth and width etc..

BTW, I had a conversation with a very experienced tube designer and he told me that the reason that tubes are inherently better than solid state is their dynamic linearity and behavior into different/changing loads.  I have to say that my experience bears this out.  Really good tube designs can sound identical to SS in almost every aspect.  I'm not a big believer in the so-called "tube" sound.  This phenomenon is a result I believe of inferior tubes or designs that limit the transient power current to the tubes, or some poor transformers or coupling caps.  If these things are dealt with properly, the imaging depth and "3-D realness" factor of tubes is just somehow better, without any sacrifice in dynamics or HF/Bass extension.  The real disadvantage of tubes is of course their life-span, but also the mechanical effects, which is not an issue in SS devices.

It's a shame that high-performance tubes are really a thing of the past.

Steve N.

[*Scotty*]

doug s said

according to published specs, the distortion on my tubed preamp is 0.15% - low enuff for me...
speakers have distortion so many orders of magnitude higher than even tubed gear, it seems any decently designed component - tubed or solid state, shouldn't have distortion as its stumbling block...

   I think doug s has been misinformed as to the magnitude of distortion present in loudspeakers. Here is a link to a website with independent distortion measurements of a large number loudspeaker drivers   http://www.zaphaudio.com/  and a link to a THD to dB conversion calculator  http://www.sengpielaudio.com/calculator-thd.htm   Some drivers are a little worse, some are better in distortion measurements than doug s preamp.
Scotty

[doug s.]

what am i missing?  if a speaker is 100db-efficient, & has a frequency response of +/-1db from 20hz to 20khz - an extremely excellent figure, imo (does any speaker meet this spec?) - isn't this, on its face, 1% distortion?

doug s.

doug s said

according to published specs, the distortion on my tubed preamp is 0.15% - low enuff for me...
speakers have distortion so many orders of magnitude higher than even tubed gear, it seems any decently designed component - tubed or solid state, shouldn't have distortion as its stumbling block...

   I think doug s has been misinformed as to the magnitude of distortion present in loudspeakers. Here is a link to a website with independent distortion measurements of a large number loudspeaker drivers   http://www.zaphaudio.com/  and a link to a THD to dB conversion calculator  http://www.sengpielaudio.com/calculator-thd.htm   Some drivers are a little worse, some are better in distortion measurements than doug s preamp.
Scotty

[JohnR]

Frequency response and distortion are different parameters.

I had no idea Zaph's site had so much information on it! Looking at the midbass plots, harmonic distortion is generally 40 to 60dB down, which is 0.1 to 1% distortion. Presumably that will go up at higher levels, and down at lower levels. If I understand right he's set the levels to produce 96dB at 1 meter.