Flacs are inferior to wavs?

If you believe that PC activity can cause fidelity differences, then two files located in two different parts of the hard disk can then result in different fidelity because the operating system acts differently when fetching one vs the other. Therefore if one takes a .wav file, makes a flac file out of it and then back to second .wav file, even though the two .wav files are identical, the difference would be there.

That is the theory. But the theory does not support their argument. There is no telling at all which one would sound better. For all one knows, the converted one would. The activities of the OS are very chaotic and not predictable at all.

The second fly in the ointment is that if one is to believe the above, then none of their tests are valid. Reason is that if the mere movement of a file from one place to another makes the sound different, who is to say all their other tests where they ripped files differently and such wouldn't do that? Worse yet, a playlist of music would have its fidelity change all the time since by definition different tracks are in different spots on disk drive.

I read the original article and do not accept that as proper level of detail to back their statements. I plan to write a much more detailed response at some point in the future. For now, I would say it is one of the saddest affairs when it comes to high-end audio. Wearing the artificial cloth of objectivity is worse than not wearing anything at all in this case.

Then don't grant them credibility. You've correctly identified the obvious. You really don't need to go looking for every crackpot detail at that point.


They can't. At least not for any reason inherent and specific to them having gone through a lossless conversion process. How about I repeat their methodology of ripping cd to flac and back to wav with various programs. However, I'll then burn them each to a separate cd and send them to TAS for them to repeat their listening tests as before. Except I won't tell them which is which. Any takers on a 100:1 bet that they can't repeat their results?

This series of articles is barely worth a skimming. Whatever this is (ie junk designed to attract or maintain a reader base and perhaps satisfy a key advertiser or two), it isn't research. It isn't objective. It's barely journalism.

What's more sad than someone writing this junk is that some people will actually believe it.

Actually no (the drive however applies error correction). But that is not the issue. In my description I assumed the file is read 100% reliably in both cases.


It does do that.


True there. But still unrelated to the point I made .


Indeed it could lead to pretty hard crashes in no time. Again, unrelated to the fidelity factor in play here.


Not quite. In the configuration the Authors tested, digital audio was extracted from the PC. The digital stream conveys the audio samples and timing for them. These *two* factors determine the analog waveform from the external DAC they used.

We have already agreed that the PCM audio samples are the same in the two cases. It is therefor the timing accuracy which could cause, in theory, some difference.

The accuracy of the timing samples can vary based on PC activity. It is a clock and like any other clock, can be impacted by environmental conditions. Reasons are complex and I won't get into them here. But for now, just take my word that it can happen. Once there, then one can get differing analog output from the DAC (see http://www.madronadigital.com/Librar...dioJitter.html).

Nah, not even close. For example, the one by that lady about blind tests being complete bunk was far sadder (or hilariously entertaining, depending on ones view of such things).




Amir, if you go to Axpona, the first beer in on me (the next three are on you - new math).

cheers,

AJ

..the problem is audibility...and ability to demonstrate it.

Anyone can claim to bend spoons and run 8 second 100m dashes in their own backyards (yeah, I know, big yard).
It's what happens when the cameras and lights go on in front of the audience...

cheers,

AJ

That's right. That is why I said what they claim is a house of cards. If such slight variations change the outcome then no other AB test is going to be valid. After all, those bits are sitting in separate files also.

It is clearly possible for bit identical files to sound different if there are other problems with the process of moving audio data from where it is stored to the output terminals of whatever DAC is being used (in the PC, in an external converter, in an AVR or headphone amplifier, etc.)

Back in the day when we played MP3 files and operated audio interfaces with PCs with a few megs of RAM, hard drives that ran only a few dozen times faster than floppies, and had processors with only a few MIPs of processing power, there were lots of problems.

It is still possible to garbage up a PC with spyware, viruses, trojan horses, gegaws, games, acessories and toys to the extent that it can't do audio cleanly or a few unfortunate coincidences push the whole house of cards over the edge.

It no longer takes rocket science to avoid these difficulties.

The most probable cause anything questionable or remarkable in a TAS, Stereophile, or other golden-ear-press ragazines, blogs, and other such anti-technical spew is the author's and editor's unwillingness to use good design principles in their alleged tests and evaluations.

Bingo!

As a rule, the TAS (and as far as it goes Sterephile's) crew scrupulously avoid doing proper technical experiments.

Any links to the first and second articles?

It can happen any place or places in the real time data path for providing the analog audio to the amplifier and speaker for listening.


In fact aside from incompetence, ignorance, and accidents, it rarely if ever happens to an extent that it is audible these days.

And that's the key - whenever someone starts trying to lead down this path ask for the reliable listening tests that they have done (DBTs - ABX tests) that support their claims. That pretty well shoos the little boys out of the way!

However, there is a lot of money in pretending that this is a clear and present danger in these days and that necessarily spending lots of money (with the right people) is the only way to feel that you have avoided it.

It is possible to write and publish an unending stream of debates and earnest technical-sounding articles about it, as well.

wav --> compute MD5 hash
wav --> flac
flac --> wav
wav --> compute MD5 hash

end of thread.

I wonder how they came up with their "Sound scores". Anybody have links to the other parts of this series?

I keep my music on a striped disk set - amazing I do not hear quality changes when the bits not only come off different parts of the disk, but off a different disk . . .

No wonder I quit subscribing to these magazines. How can they expect us to accept the equipment reviews when they publish what appears to be drivel to me!

Edit: I've taken a music file - converted to flac - alac - wav - alac - flac - aaif - alac- wav in various combinations. In all cases I end up with a perfect copy of what I started with (at least the unix diff command tells me the binary files are identical)

Good one. Got a laugh out of that. I would be willing to bet that this is out there:

Tracks on / near the end of the CD sound better because you have faster linear track read with less lens movement.

These two individuals (they aren't authors) are completely mistaken in their understanding.

I believe the FLAC file gets unwrapped, the wav file extracted and then played back. At the point of unwrap you have an identical wav file to the one that wasn't compressed. It's math (like the MD5 hash). It is what it is.

In the days where you can operate on 60MB of data in a mere 1 to 1.5 seconds on clean, properly running computer a 14MB file that is 7 minutes long in audio isn't any form of issue.

I'm surprised they didn't test Registered ECC, Non Registered ECC, Non ECC RAM

from the part 3 article

"To this end we performed 5 or 10
consecutive FLAC-to-WAV conversions
using dBPA on each computer in System
1 and System 2, respectively. Given the
potential for controversy, depending on
our results, we repeated our listening
tests in three independent sessions with
3 listeners in each. We consistently found
that there was a cumulative stepwise
decrease in overall sound quality (audible
as a decrease in image height, stage depth,
clarity, and rhythm and pace). We also
found that the overall effect was larger
for files made on the computer of System
1 (32-bit Windows Vista, 4 gigs RAM,
2.66GHz CPU) compared to System 2’s
computer (64-bit Windows 7, 8 gigs RAM,
3.0GHz CPU)."

This is just flat out ludicrous. At least they thought it might be controversial.
You can go from flac to wav and back and forth as many times as you want,
on any computer you want.
The result will always be the same (unless I suppose they use a program
that does not convert properly).
Maybe they would be happy if we compressed music using zip instead of flac? Jeez!

It is possible that playback software makes a difference, especially if they are doing upsampling.
But to say that the number of times a file had been converted from one format to another could possible make a difference when the resulting files were always identical is just crazy.

I think now the purpose of the article was not to compare flac to wav - the purpose was to generate letters to the editor. . . so the magazine would seem more important and generate more advertizing revenue.

Actually I read Stereophile to see results of measurements they do. If some article is not complemented by objective test data, I consider it just entertainment, rather than seriouse reading.

But considering TAS subscribers base, these articles touch the string in their souls. This is all what matters for magazine publisher.

Maybe that is in part 4 Or maybe they will compare the results of converting between formats on a Mac vs. PC vs. Linux

You would think by now they would have realized the obvious. Their listening tests are flawed.

Edit: at least it was an entertaining read - thanks for sharing

A big problem is that such magazines do have good writers that promote snake oil convincingly. TAS is notorious for this and Stereophile is little better, my apologies to Sal.

It isn't end of the thread as I explained earlier. Nor is it something the authors debate.

Please don't throw stereophile under the bus. They refused to publish the article: http://www.whatsbestforum.com/showth...8993#post78993

"The authors wrote an article on blind testing of digital systems for Stereophile in the mid-1980s. They offered this computer audio article to Stereophile but I felt there were some technical issues that needed to be addressed before the article could be published. Rather than address those issues, the authors withdrew it and offered it to TAS, where it appears to have been published as is.

John Atkinson
Editor, Stereophile"


Stereophile is also one of the largest libraries of *objective* measurement data for audio equipment. If that is not pro-science/pro-technical, I don't know what is.

What is anti-technology is the non-sense that in this day and age people hear differences in PC audio files because of performance issues as you claimed here and elsewhere. You are taking the listeners for idiots in making that statement.

The vast majority of Stereophile's content is information developed via listening evaluations. Most of their articles provide other than the results no measurements.

I congratulated John for his good sense in stepping back from this article. Doing the right thing once does not cancel out or balancve the thousands of misapprehensions that have been published in the past.

It's a classic old time medicine show approach, mixing in a few reliable facts to make the whole sorry potpourri look solid, when it is not.

Amir, you appear to be trying to sell Stereophile's Kool Aid, which is a modicum of reliable facts combined with the vast majority of it's prose which is not based on reliable testing.


Here's what I wrote:


Where did I say that?

So do I. But no way do the minority of magazine content which is based on technical tests that are probably reliable balance out the overwhelming majority content, which is just subjective personal opinion.


Good call!


Getting your articles mentioned by other publications and bloggers is just another marketing technique for them.

It doesn't matter what percentage of the print magazine is the objective data. As I said, what is there is worth its weight in gold for us objective crowd. Your insistent attitude to put them the whole magazine is tiring and anti-science.


Which would then been good to note instead of throwing a rock at them when no one was talking about Stereophile.


Go start a rant thread against stereophile but don't pollute this thread with it when they had only taken a positive step in this regard.


I did not sell you on anything but the *objective portion of stereophile.* That you have no use for it and want to put the magazine down constantly comes from your battles with John. That is not my concern. When I want measurement data that is there for the whole world to read for free, I go to stereophile magazine. There is no other source whatsoever. They also have extensive tutorials on technology and operational architecture of equipment they review. These are unique assets to be cherished by objective audio fans. To argue otherwise is the height of emotional need taking over objective evaluation.


You said it is "clearly possible" for bit identical files to sound different due to performance issues due to lack of MIPs, etc. That is the worst excuse I want to give someone who thinks they thought one file sounded better than another. The audible effects of the things you mention is distinct and completely different than thinking one file had better imaging, better detail, etc.

I'm very sorry to confuse you with a true statement of well-known fact, Amir. ;-)


It's not the only reason I gave. I do understand Amir that in your book reliable relevant generally accepted facts are just sorry excuses for the message you want to send.



Amir, I attribute those things to the authors taking your advice as recently posted here, and avoiding doing reliable listening tests.

The latter is all you should have said. The rest in this post and others here are a mix of folklore and emotional need to fight other wars which are a distraction.

Have you even read the three part article? I have. Until you do, you don't even have business complaining about the one bit I gave you credit above.

I'll say! The articles read like crap.