Good point! IIRC, one of the best systems I heard used galvanic (transformer) isolation of the supplies and opto-coupling of the digital lines into the DAC.
The Absolute Sound (magazine) take on many aspects of computer assisted music reprodu
- Thread starter rbbert
- Start date
garylkoh
WBF Technical Expert (Speakers & Audio Equipment)
Absolutely. With the music server we built for the Pacific NorthWest Audio Society, the motherboard Toslink connector into the Weiss Minerva sounded far better than the coaxial S/PDIF connection. I believed that it was due to the galvanic isolation achieved with the non-metallic Toslink cable.
I certainly do mean it. I'm an academic biologist. Show me some positions which are 'middle' -- i.e., one in which the evidence is just as good on one side, as the other. Certainly the thing that the lay public considers 'controversial' in science -- evolution, anthropogenic climate effects, and, to bring it on home, audibility of most audiophile nostrums -- are not really controversial in science (including psychoacoustics).
Last edited:
..which, of course, makes no sense. One doesn't 'burn audio CDs' from FLAC files. They are of course expanded to .wav first. And it should make absolutely no audible difference unless your system is broken.
And I strongly suspect that in properly-functioning instance the audible differences are wholly imaginary.
It's amusing to see how far you guys (it's almost all guys) are willing to burrow down the rabbit hole.
It all depends whether you consider the world to either be made up of a series of completely independent black boxes, or at some level it all exists as a continuum. And your language indicates that you take a "digital" view of the first scenario: either elements, components of an audio system in this case, are truly behaving as lone entities; or they are defective, it's either one thing or the other. I, and others in the audio world, take the view that once you link these elements by some means, or they are in close proximity, that then at some level non intended interactions result, and in the case here of audio systems, audible differences will be heard.
Frank
Yes, in real time by the burning program, and in my experience this will not uncommonly lead to errors at the highest burner speed. Not a subjective phenomenon at all, as evidenced by the different data found on the disc. Just try it if you don't believe me.
Almost all medical and related research (as well as psychoacoustic) is considered "real" or "positive" at a p-value of 0.05, or a 19 in 20 chance of being real, and that's only for the specific conditions of that study or experiment. It turns out to be fairly common for other tests/studies turn out non-confirmatory or even contradictory data when conditions may not be identical, even when not recognized as such. Since study conditions in any area of science are often not replicated in the real world, one is left in a middle ground.
To give a couple of simple examples with wide implications? Coronary angioplasty vs. medical therapy for coronary artery disease, arthroscopic knee surgery vs. non-specific therapy for osteoarthritis
I asked my friend and his memory matches mine that on the GP PC we saw many spurs in the -60 dBFS range with some up to -40 dBFS, and a noise floor that was not only higher but had numerous "humps" in it indicating deterministic noise was added. The dedicated server had spurs better than 80 dB down with a couple (clock-related) pushing -70 dBFS. We also mesaured the ambient noise outside the boxes, and of course you could hear the fans and drives whirring in the GP machine, but we were most interested in the analog output. We did compare the two machines side-by-side, and you could clearly hear the difference in noise floor with small signals, and the additional spurs were discernable with test tones. With musical input it was much harder to tell, of course, but we did far less of that. These were not blind tests, and being a hairy-knuckled engineer you could assert we have ears of clay. The measurements were pretty clear, however. Note that, as stated earlier, our conclusion was that most of the noise was coupled into the analog outputs via the power supplies and various EMI/RFI paths, not that the source data (bits) had anything to do with it (other than jitter, of course).
I sill cannot understand how two bit-perfect files could have any influence, but there are plenty of other potential causes for variance.
I'll let you have the last laugh - Don
My experience also is that dedicated servers are often a much better starting place than a consumer PC. Also, the really special-purpose devices such as the Auraliti offer further advantages, meaning there's not much inside them and they're designed from the beginning for audio purposes.
garylkoh
WBF Technical Expert (Speakers & Audio Equipment)
Either their system is more transparent than mine, or their WAV > FLAC > WAV conversion is less transparent because I sure can't hear a difference no matter how many times I convert WAV > FLAC > WAV > FLAC > WAV. Heck, I can't even hear the difference between WAV and FLAC.
I wish it was just that. But they go miles past that.
Example: they claim that they did 10 passes of flac to wav and back. And in that in blind testing, they found every step to degrade the sound further and further! Unbelievable. I swear they must have done these tests sighted.
Did they bother to do a binary comparison between the files? Nope. They relied on said blind tests instead. Why, why, why? If tools and mathematics can prove something to be inaudible, let's try them for heaven's sake. How could we even remotely appear objective otherwise?
They claim that they difference must be due to meta-data/header in the file. Really? How did that change over 10 tries?
The basic premise that *playing* a flac file may sound different than wav at least has a basis for an argument. But the stuff they wrote in this article just boggles the mind. And you know me, I am the best friend they could have in any of this!
Their core proof point for all of this is these point system they have cooked up on the results of the tests. This sounded 25 points lower, that sounded 10 points higher. Really? This is that mathematical that we can make comparisons like this?
I think one has to start wondering about their motivations behind writing and publishing this series of articles. Doing the research and testing I can easily understand, but when they started to get results like this they should really have taken a step back and begun to systematically investigate their methods (specifically their playback chain and techniques).
Worse thing about this is that they talk as if they are scientist. When they present the first surprising result they follow with, "let's quantify this." And their "quantification" is to claim the blind test rating was 150 vs 120 or some other such non-sense. This is quantification?
Quantification would be taking some measurements to back the listening tests. They try to prove their point with the point itself. We heard differences, and the proof is that we heard differences of this amount and that amount.
Quantification would be taking some measurements to back the listening tests. They try to prove their point with the point itself. We heard differences, and the proof is that we heard differences of this amount and that amount.
Bruce B
WBF Founding Member, Pro Audio Production Member
You are a nice person Gary .. I will not attempt to out-nice you. This article is badly flawed ... PERIOD!
I certainly agree to this. Their system must be broken. That's the only way.
Bruce, one of their arguments about why they may be right (about FLAC) is that professional
audio workstations don't use FLAC.
Similar threads
