Baffled about computer power

[microstrip]

I don't understand your question. If you are asking if I have also done listening tests, and spent far more with these systems than the measurement project, the answer is yes. I have lived with these systems for 2-3 years in this architecture (and more than a decade using Pro Audio cards). We have three music servers at work in different rooms/systems and have them connected using normal USB, the two async interfaces and multiple speaker/amp configs. I have spent hours doing AB comparisons. Those subjective results point to the async USB interface improving the performance of any external DAC/Processor that I connected them to, sans one. They also always outperformed non-async USB subjectively.

I have also documented my Berkeley vs Audiophilleo here: http://www.whatsbestforum.com/showthread.php?4160-Review-Berkeley-Audio-Alpha-USB

I don't push this data because as I say in that thread, everything I hear may be faulty. The effects of placebo are far stronger than the differences I hear. For that reason, I present the measurements because they are reliable and concrete in what they show.

We may also differ in priorities. To me the *content* of music is priority 1. Its fidelity, as shocking as it may be to many of you, is priority 2. I get far more enjoyment out of my new car stereo that lets me play my large music library from flash drive where I can listen to a ton of varied music as my mood mandates. In contrast, the CD is boring in that I wind up listening to the same CD over and over again as I drive long distances. Convenience for me is such a huge factor that once I adopted music servers, there was no going back. For that reason, you don't see me with a stash of high-end CD players to compare things to. I think they are obsoleted and dead, just not buried . They are the rotary phone of music systems. We absolutely know how to get superb performance out of a PC and the measurements show that in spades. As do the superb fidelity that I hear. At shows like high-end area of CES, 9 out of 10 digital systems I hear are based on music servers.

So don't look to me spending a ton of time worrying about what $$$ stand-alone players are like. It just isn't a priority for me either based on usage or what they think they can accomplish technically. All of this is of course opinion. The data was not .

For someone who seemed not to understand the question you answered pretty well - I can now understand much better your opinions and weight your points accordingly. Your personnel weight of convenience and interests are quite different from those of many audiophiles, experience and opinions of audiophiles such as me are of little use for you. Your use of shows as an example only evidences how far you seem to be from high-end and some audiophiles - we often read that 9 out of 10 exhibitors have poor sound, most of those who have great sound rely on analog or music servers using exceptional recordings or finely tuned systems.
BTW, you seemed to misunderstand the point of the CD transport importance in this affair. The question is not the rotary phone aspect, it is trying to understand the why's of their performance coupled with matching DACs of the same brand and seeing if something can be learned from it that can be used to improve PC music player sound quality. Since you consider it superb, it is not a question anymore.

 

[jkeny]

Amir, I'm not saying that the test was not properly conducted or that it was not sensitive enough to show a PARTICULAR difference between two digital delivery systems & indeed I applaud you for it & did so at the time you published it. What I am saying is that it shows just one PARTICULAR aspect where PC audio measures better than a transport - it is not the full picture. To draw a conclusion from it regarding transports Vs PC audio is premature. You have since established that you have also listened & formed your conclusions based on subject comparisons of these two digital delivery methods.

My experience is that it takes a very good PC audio system to beat the best transport systems available & a lot of the reason for this is related to noise issues stemming from the PC systems. There are a lot more noise sources within PC systems that transport systems do not have to contend with.
It may well be that if you had measured noise you would have found a very different picture between the two digital delivery systems.
As Steve stated & Opus agreed, common mode noise is one such issue.

As John Swenson stated isolation in his XMOS system removed USB cable variations but still left PS supply as a variable in that he could hear the audible influence of different PS driving his PC.
I concur with his experiences regarding PC PS, as it is mine also.
This to me signifies other elements at play that may not be ameliorated by isolation alone for whatever reason & have yet to be identified.

 

[jkeny]

... But you guys are doing such a poor job of making your case, and providing details that I feel no motivation to go beyond tests that so clearly show what we want to see.

Just taking that last bit of your statement "I feel no motivation to go beyond tests that so clearly show what we want to see" could easily be read to mean - as the test shows what you want you therefore do not want to do any further testing - but I presume it shouldn't be read in this way?

 

[amirm]

For someone who seemed not to understand the question you answered pretty well - I can now understand much better your opinions and weight your points accordingly. Your personnel weight of convenience and interests are quite different from those of many audiophiles, experience and opinions of audiophiles such as me are of little use for you. Your use of shows as an example only evidences how far you seem to be from high-end and some audiophiles - we often read that 9 out of 10 exhibitors have poor sound, most of those who have great sound rely on analog or music servers using exceptional recordings or finely tuned systems.

Just as I guessed what you were asking, I predicted this answer would come from a mile away. I say that music is paramount to me as art, ergo it must be that fidelity is not important to me. No, it doesn't work that way. True enough, I have watched wonderful movies on a Hi8 tape on Transcontinental flights. But that does not at all translate into me still searching for VHS tapes for movies. To me, the day that I care about fidelity more than the creating aspects of a recording, is the day that I think I have lost my way as to why I am even in this hobby. I like to hear if you put quality of recording ahead of the artistic aspects.

You say 9 out of 10 suites at shows don't sound. Since those companies are all by definition "high-end" would you like me to conclude that you don't consider high-end, high-end? How is it that if I blend into that crowd by using music servers, I am not into high-end, or as you say what members of this forum care about??? You guys have created a special place for yourself that is a far more exclusive club than 90% of the high-end audio companies? Exactly which companies fall in that category?

This is all news to me that using a music server and caring about the art above and beyond all else automatically excluded me from the club. I will leave it to Steve to decide to ban me from the forum or not.

BTW, you seemed to misunderstand the point of the CD transport importance in this affair. The question is not the rotary phone aspect, it is trying to understand the why's of their performance coupled with matching DACs of the same brand and seeing if something can be learned from it that can be used to improve PC music player sound quality. Since you consider it superb, it is not a question anymore.

That concept is completely known. It has been known for a decade or two. PCs are actually used to create the content that you play. If PCs are so awful and still have to figure out how to sound good, how is it that the talent heard and approved the content on said PC/Mac workstation? Is Bruce delusional when he uses his workstation to proof and master digital content for distribution?

Clearly that logic is flawed at multiple levels. No doubt you know that companies like Meridian actually copy the PC architecture by using a CD-ROM (data) drive rather traditional CD transport. Do we take them out of the high-end club for that???

All of this said, yes, I am different than most of you. As an engineer and guy who has manged to head design of many systems, who has participated in hundreds of listening tests, and has read more papers on audio science than they teach in two graduate degrees, I don't think about audio like you guys do. Everything I say or do, goes through a process of analysis and judgement applied to it that is different that someone who treats these things as black boxes. I cannot suspend science and wish for new measurements that I can't articulate. You all can. And that is OK. Just don't throw me out of the audiophile club because I don't put aside four decades of experience on the other side of the fence on this thing .

 

[amirm]

Just taking that last bit of your statement "I feel no motivation to go beyond tests that so clearly show what we want to see" could easily be read to mean - as the test shows what you want you therefore do not want to do any further testing - but I presume it shouldn't be read in this way?

You should read it exactly as I wrote it: "But you guys are doing such a poor job of making your case, and providing details that I feel no motivation to go beyond tests that so clearly show what we want to see."

See what I bolded? If I tell you that a sports car can be characterized with 0 to 60 times, braking distance and cornering forces, and that makes your car look bad, you don't come out and say, "well those are the tests that show what you want so they don't count." And when asked what test you suggest, you have nothing to offer. That's a problem.

 

[opus111]

For those more interested in understanding common-mode noise than playing psychological games, here's some helpful measurement data from the guys over at http://www.exasound.com.

Firstly they show how their DAC looks when playing digital silence and with galvanic isolation from the PC:



Then here's the plot when there's no galvanic isolation - again, no audio signal is present for this measurement:



The take away from these plots is that the noise is largely ultrasonic - OOB in other words. Thus if you make an FFT measurement in the audio band only you'd miss the huge lump (50dB higher than the LF spot noise) around 100kHz. Since with galvanic isolation there's capacitive coupling between the DAC and PC grounds (around 17pF according to Exadevices, the spec on the USB interface they use) the noise will tend to rise with increasing frequency.

 

[opus111]

Perhaps you didn't see it, but I did analyse tape over on one of my 'Frugal Audiophile' threads and yes, it too suffers from noise modulation.

Perhaps you could explain this to me but why does there need to be a competition between analog and digital? I am of the view that digital has the potential to sound better than analog, but this potential has yet to be realized in practice. One reason it can sound better is that digital recording does not need to introduce modulation noise as analog tape inherently does.

 

[dallasjustice]

Opus,
What would the theory that would explain audibility of the ultrasonics shown in the second measurement? Quantization error folding over into the audio band? If so, wouldn't that type of distortion be measurable with an FFT?

For those more interested in understanding common-mode noise than playing psychological games, here's some helpful measurement data from the guys over at http://www.exasound.com.

Firstly they show how their DAC looks when playing digital silence and with galvanic isolation from the PC:

View attachment 10265

Then here's the plot when there's no galvanic isolation - again, no audio signal is present for this measurement:

View attachment 10266

The take away from these plots is that the noise is largely ultrasonic - OOB in other words. Thus if you make an FFT measurement in the audio band only you'd miss the huge lump (50dB higher than the LF spot noise) around 100kHz. Since with galvanic isolation there's capacitive coupling between the DAC and PC grounds (around 17pF according to Exadevices, the spec on the USB interface they use) the noise will tend to rise with increasing frequency.

 

[opus111]

What would the theory that would explain audibility of the ultrasonics shown in the second measurement?

Excellent question.

No active electronics circuit is perfectly linear, they all suffer from some degree or other from non-linearity. Whereas the most popular way to demonstrate the non-linearity of a piece of electronics is a THD measurement, this turns out not to be the most relevant where SQ is concerned. That's because music is not a single sine wave, but a multitude of them. When there are too many tones present to even count them, intermodulation distortion dominates over harmonic distortion. Intermodulation distortion (IMD for short) is what happens when there's non-linearity and more than one tone is present - the tones beat together creating sum-and-difference additional tones. But with so many tones the distortion appears as general grunginess - that's what I call noise modulation.

Notice that with noise modulation, with no signal present there's also no noise. That's because there's nothing to beat with - so traditional noise measurements won't show any problem with the noise floor. But as soon as you play music through the noise modulation appears as the ultrasonic energy combines non-linearly with the music to produce audible grunge.

I say the grunge is audible, but often its not perceived directly - which is why the descriptive terms are most often 'loss of dynamics' rather than 'presence of grunge'.

 

[Julf]

I studied electronics myself to degree level, maths was fairly intensive in my second year. The textbook was Erwin Kreyszig, Advanced Engineering Mathematics. I recall nothing from there - would you be more specific?

Have to admit I can't remember if the Taylor series was covered in maths or electronics. It's been a while...

A link would be cool.

www.ugcs.caltech.edu/~kwalsh/npr_uniform.pdf? is an example (from 1999) of the two-tone test, where the subsequent analysis is performed using a comb filter. It appears to be a further development of the work published in the (1977) WW article.

 

[amirm]

This is one of the stimulating and fun aspects of having discussion with objectivists - they're almost always crap listeners.

Love to see your data on that. Here is mine. At Microsoft, we routinely tested our audiophile group (hundreds) to get large scale test results beyond our expert listeners. With very rare exceptions, they were all just like general public, and could not hear small non-linear distortions. These were in double blind tests where the expert listeners which in most cases were not audiophiles could easily detect the distortions down to exact note/sub-second accuracy. Yet the audiophiles could not hear them at all let alone down to one note.

In one test, we were evaluating whether we could achieve "perceptual lossless" encoding. This is a lossy encoder that is told to use its psychoacoustics model to only discard what it absolutely thinks is inaudible. I managed in double blind tests to find degradation in a clip that none of the audiophiles had found. I guess this makes me a subjectivist???

In yet another test on AVS Forum, clips were put forward blindly for people to evaluate. By accident, two of the clips were the same. One of the people who did the test was a professional which recorded/mixed movie sound for a living and had done a number of blockbusters. He was a professional subjectivist if you want to call him that. He thought the identical clips were different! I can give you the link if you like. Again, I managed to find the duplication.

Again, in all of the cases, the people with crap ears actually heard the differences that existed (or lack hereof) and it was the self-selected audiophile group that did not, under exactly blind testing circumstances.

Now, occasionally an audiophile will have the same ears as an expert listeners but so do people off the street. I was neither. I could not hear any artifacts until I trained myself. Then it became trivial. Harman put this test forward on whether we can tell if a signal is EQed with a single band. Initially I could not tell much past 4-5 levels and neither could anyone else. With some practice though, I quickly rose up above average level. In a group test, Sean ran the test and I managed to do better than everyone else in the group but Sean beat me easily. I think I could get better though if I kept testing and approach their internal test group. Here I was as an audiophile for decades but still needed training to hear distortions.

So it is clear, I have also been known to be a crap listener on more than one occasion not being able to tell identical sources were the same and so on.

Let's see if you have concrete data to back that statement. Otherwise, unfortunately the reality wins: when it comes to non-linear distortions, we are essentially all "crap" listeners. If we were not, we would not fail so many blind tests where we know from measurement there are differences. This is why I always ask people if they have ever objectively tested their hearing ability before assuming they can hear small distortions.

 

[jkeny]

It's a pithy comeback, John, but there's no truth in it. Objectivists deliver positive results in listening tests all the time. Just give them something that can clearly be heard.

Tim

Tim,
You fall for it every time - your expectation bias that is!

 

[amirm]

Here is the other reason we are crap listeners: we don't agree on the same thing!!! Why is it that people buy so many different amps, CD players, cables, etc? Definition of a good hearing would be to hear the truth. How could there be hundreds and thousands of versions of truth? Shouldn't we all go to the same show, hear the same gear, and come back with identical assessment as audiophiles if our ears are not crap? Put aside speakers where the difference is apparent and tell me why three audiophiles picked at random will not have the same view of the gear with small distortion.

We have some really dark secrets here . Best left unspoken than to brag about it the other way around.

 

[Groucho]

Here is the other reason we are crap listeners: we don't agree on the same thing!!! Why is it that people buy so many different amps, CD players, cables, etc? Definition of a good hearing would be to hear the truth. How could there be hundreds and thousands of versions of truth? Shouldn't we all go to the same show, hear the same gear, and come back with identical assessment as audiophiles if our ears are not crap? Put aside speakers where the difference is apparent and tell me why three audiophiles picked at random will not have the same view of the gear with small distortion.

It could be because they sound different, but there is virtually no proof of it. Far more likely reasons are:
- prejudices and biases towards equipment that appeals to our sense of aesthetics/nostalgia/view of what is 'correct'/need to tinker/
- propaganda
- 'marketing'

In my case, I am happy to admit that I love to tinker. I have a view of what is 'correct' (the ultra-modern approach of active speakers, DSP, digital source, plus not-so-modern approach of uncompromisingly large speakers), plus a bit of nostalgia for old-fashioned gear. Aesthetics is a little way behind this, but I will avoid anything that looks nasty without a good reason for it. As for the listening part, I really do believe in design and measurements and that everything will fall into place if this is done well. So in effect I am saying that if I don't like the sound of gear that is objectively good, then my hearing needs educating rather than the other way round - luckily this doesn't seem to happen; I find that systems put together without 'fudge factors' (and I would also count analogue sources, valves, passive speakers etc. as effects boxes which fudge artificial 'musicality' into the equation) sound good.

 

[amirm]

It could be because they sound different, but there is virtually no proof of it.

Let's say it is a given that they sound different. If audiophile ears are not crap, and tell the truth about fidelity, why is it that all the votes don't gravitate toward one gear? Clearly all but one must be wrong if we are a good judge of true fidelity.

So one of two things is wrong about the above:

1. The equipment does not sound different and we are imagining the difference for reasons you mentioned. Since we can be different in what we imagine, that would explain the differences in votes.

2. The equipment does sound different and our ears are crap. We are unable to pick the good from bad.

I cannot escape this duality just the same. It is for this reason that I rely on a foundation of how technology works, what research exists about that and how we hear, and measurements. All of this helps to tilt my opinion closer to the truth. It doesn't guarantee it of course but increases the odds. Merely being an audiophile as stated does little here.

 

[Empirical]

Amir - This power supply post on Asylum was mine.

We should get together, afterall I just live in central Oregon, about 5 hours from you. I'm located at the premier resort in Central Oregon, Black Butte Ranch, so you can even come here for a vacation at the same time. Great golf courses, rated in the top 100 by Golf magazine:

http://www.blackbutteranch.com

I have equipment and techniques that I use for directly measuring jitter here that I believe you would be interested in. I also have a system resolving enough to easily hear differences in .wav and AIFF files as well as ALAC and FLAC files.

You could bring your front-end DAC and USB converter etc.. for comparison.

I also have a lot of top-end equipment that I have used at shows here that I'm considering selling.

BTW, I contacted the Microsoft audio group before Win8 was launched and proposed a new infrastructure for audio that would allow a whole new industry to be created with higher quality playback for every audio and video system from top to bottom. They seemed to like it, but dropped it because Win8 was consuming them. We could talk about this as well.

Email me if you are interested.

Steve N.
Empirical Audio

 

[mep]

Let's say it is a given that they sound different. If audiophile ears are not crap, and tell the truth about fidelity, why is it that all the votes don't gravitate toward one gear? Clearly all but one must be wrong if we are a good judge of true fidelity.

Amir-I think it's more complex than saying audiophile ears are crap. At the present state of the art, there is no "absolute fidelity" that any manufacturer can lay claim to. Because there are no perfect electronics and speakers, everyone gravitates to either their version of what they think the truth is or maybe just what sound pleases their ears. I started a thread a long time ago here called "It's All a Preference" and the things I wrote then I still believe to be true. We choose gear because we prefer the way it sounds compared to other gear. And I'm sure in some cases people buy gear based on the measurements because they trust the measurements written on paper more than they trust their lying ears.

So one of two things is wrong about the above:

1. The equipment does not sound different and we are imagining the difference for reasons you mentioned. Since we can be different in what we imagine, that would explain the differences in votes.

A case could be made that bit perfect files regardless of source should sound the same when they hit the D/A converter. However, since there are a zillion D/A converters on the market and they can certainly have different topologies, parts, and measure differently, people are going to choose their digital setups based on how it sounds to them (there is that preference word again).

2. The equipment does sound different and our ears are crap. We are unable to pick the good from bad.

I'm voting that equipment does sound different (specially gear like tables, arms, cartridges, preamps, power amps, D/A converters, and speakers) and that our ears aren't crap. It's not that we are unable to pick the good from the bad, we pick the sound that we prefer.

 

[jkeny]

Hi Steve,
Welcome to the forum.

Don't know if you read the entire thread but I hope you found something of interest here?

In relation to your experience with different PS powering the Mac Mini, have you any further datapoints to add to this observation in the form of test results? I know you have looked into CM noise, anything you can share?

I found John Swenson's observations about XMOS isolation interesting, any comments?

 

[microstrip]

Let's say it is a given that they sound different. If audiophile ears are not crap, and tell the truth about fidelity, why is it that all the votes don't gravitate toward one gear? Clearly all but one must be wrong if we are a good judge of true fidelity.

So one of two things is wrong about the above:

1. The equipment does not sound different and we are imagining the difference for reasons you mentioned. Since we can be different in what we imagine, that would explain the differences in votes.

2. The equipment does sound different and our ears are crap. We are unable to pick the good from bad.

I cannot escape this duality just the same. It is for this reason that I rely on a foundation of how technology works, what research exists about that and how we hear, and measurements. All of this helps to tilt my opinion closer to the truth. It doesn't guarantee it of course but increases the odds. Merely being an audiophile as stated does little here.

Amir,

I am astonished that you ask these questions. It is all explained in the introductory pages of the Sound Reproduction by F. Toole - why in spite of not knowing the truth we can still be good judges of sound quality of audio systems. And the large spread of opinions is also mainly due to the intrinsic limitations of the stereo system and the instability of the stereo reproduction, that needs a large contribution of the listener to recreate the illusion. As Toole remarked "stereo, therefore, is not really a system at all but, rather, a basis for individual experimentation", relying in a few cues and a lot of experience from people.

In spite of their variation, most high-end audiophiles have similitude enough in their findings to establish a dialog about their own experiences and equipment evaluation and use with success other people experience. The system is not perfect, but works fairly well. IMHO, constructive people exploit correlations and try to get valid information from similitude of appreciations, those disliking high-end exploit the antagonisms or the human aspects not related to sound quality of the preferences, such as ownership and friendship, to discredit the others.

 

[Julf]

It is all explained in the introductory pages of the Sound Reproduction by F. Toole

I guess the Toole book is one view. Perhaps not the only one.

In spite of their variation, most high-end audiophiles have similitude enough in their findings to establish a dialog about their own experiences and equipment evaluation and use with success other people experience. The system is not perfect, but works fairly well. IMHO, constructive people exploit correlations and try to get valid information from similitude of appreciations

Isn't that precisely the description of groupthink?