AIFF Vs. WAV

[4est]

"Since testing people instead of equipment seems likely to be abhorrent or insulting (or threatening, in some cases, perhaps) to the very people who hear WAVE versus AIFF differences, I can offer no realistic solution to solving the WAVE/AIFF mystery."

 

I regret ever having posted on this thread.

 

[goldsdad]

"On the topic of different computer behavior interfering with the audio equipment through the mains or even by electromagnetic induction: highly unlikely, I am really sorry but any physics undergrad could tell you this. The iPhone or iPad you're holding in your hands or the Wi-Fi in your room cause much more disturbance than the slightly different behavior of a computer ever could (especially through electromagnetic induction). And for the case of disturbances through mains I suggest testing by making the computer do completely different things (say nothing but play music and then play a movie while downloading and playing music). Do a double-blind test to tell us if you hear a difference and report back."

 

I agree to a degree. My argument was not intended to suggest that I hear any difference between WAVE and AIFF playback. I don't. My position as a skeptic of there being differences that are physically real (certainly at an audible level) should have been pretty clear throughout this thread.

 

I raised the issue of electromagnetic noise differences due to file format-dependent processing as an example of how your proposed testing procedure may be flawed in eliminating all variables except file format. Your quote above includes "highly unlikely". Your own words allow a possibility.

 

 

[goldsdad]

Forrest, why are you taking this personally? I thought we had come to a mutual understanding of our positions when communicating outside of the forum.

I'm truly sorry for any insult that may have been inferred from my posts in this thread.

 

 

 

[4est]

Exacerbated is more like it.

 

I keep reading this thread in hopes that someone might have something new to enlighten me with on this topic. I was warned about this sort of outcome from the outset off thread by a very knowledgeable poster here. All I have ever wanted to do is supply a data point for the OP. I have never made any claims as to having great hearing, knowledge or insight on this matter. In fact, I have stated numerous times that it may just be my equipment. In my world, when someone repeatedly asks the same question, it is because they do not believe you.

 

Here in the Midwest US there is a heat wav(e), and so maybe my skin is thin. Regardless, I am weary of being the point man on a fruitless mission. I do not like this truth, and have done my best to keep my eyes closed and listen to AIFFs- only keeping a small playlist of favorites in .wav form. I am not so foolish to attempt to enter the realm of proving anything to anyone. Fact of the matter is, that I am probably the dumbest kid on this block, but I am smart enough to know that.

 

[Paul R]

 

-Paul

 

 

 

 

[goldsdad]

@ Paul & Sik

 

Regardless of whatever measurable differences there may be, surely the only one way to find out if people can aurally detect differences without expectation bias is to test the people.

 

Errrr... I'm back where I started. My train must be on a circular line.

 

 

[Paul R]

We need to have a person say "I hear a difference between this and that", have a reference for what "this and that" really is, and record the sound they are hearing to analyze it for any repeatable differences.

 

It does need to be done over a fairly wide audience, and the results to be analyzed have to be normalized or otherwise adjusted to account for differences in what is recorded. In other words, the differences between "this and that" are not absolute between systems.

 

It can be done, but there are a lot of considerations.

 

-Paul

 

 

[goldsdad]

What do you think of this rough outline of a test?

 

People can be tested without fear of losing face. In fact, they can be given a chance to prove that they have good, or even great, aural acuity.

 

Imagine a double-blind listening test as follows.

There is a set of recordings. They may have been chosen by the test subject for familiarity. For each recording, there is a WAVE file and an AIFF file such that the files contain identical audio samples of the recording.

The subject listens to a number of paired playings of each recording with no possibility of prior knowledge of whether each pair will be identical files or different files.

 

If any one test subject can, with statistical significance, indicate that paired tracks, even if of just one recording, are either identical or are different (no details of the difference need be described), then they will have demonstrated that it is possible for a human being's response to audio playback to be affected by AIFF or WAVE file format alone.

 

If a person cannot pass the test, then they don't lose face. The test equipment or conditions can be fairly held as possible reasons for that outcome.

 

 

[Paul R]

I would also suggest that the files have to be pre-prepared and just downloaded.

 

The "reward" would need to be somehow telling the user what they choose and how that ranks among other anonymous testers.

 

We might also want a control group of testers who test twice - once "blind" and once with full knowledge of what they are listening to.

 

I do think we need to record what the user hears and analyze it, but that would be best done at an audio show or someplace where a lot of people can gather.

 

-Paul

 

[Sik_Lescinovid]

Don't worry goldsdad, I am very well aware of your position

As to the matter of electromagnetic induction the "highly unlikely" was supposed to be a countermeasure to all the boring people who always mention the trivial case of everything possible happening by the laws of quantum mechanics (i.e. everything that can happen does happen). Although that is true we are talking about unimaginably small probabilities here. The likelihood of a computer (or attached audio cable for that matter) creating a strong enough variable magnetic field to allow for induction in audio equipment as near as 10cm is of such unimaginably low probability. The only reason I mentioned this possibility is to avoid having to deal with smart-asses, but it seems to have backfired (not that you're a smartass!!, just the "deal with" part); but I do agree with their being certain technical details that have to be brought into consideration when actually doing the test, the above just isn't one of those. I also fully agree on the human-testing part, but as you've stated most would not commit to such a test and thus I am prepared to settle for the abstract data. But I will also say that if there are no measurable differences (in a general sense i.e. for any system) than there can be no aural differences and thus I am also interested in just obtaining some non-aural data for the sake of doing at least something concrete.

 

[thesurfingalien]

It sounds like an ABX-test without listening to A...

 

/If a person cannot pass the test, then they don't lose face. The test equipment or conditions can be fairly held as possible reasons for that outcome./

 

Does that not make the test pretty meaningless???

 

Regards,

Peter

 

 

 

[Paul R]

Since you are so new here, why don't you tell us a little about yourself in the occupation thread?

 

http://www.computeraudiophile.com/content/Whats-your-occupation

 

 

 

[goldsdad]

@Peter

 

The test is simply to determine whether the subject is affected differently by stimulus A in contrast to stimulus B. It is not to test for the subject having an ability to correctly identify whether the stimulus is A or B.

 

The test is not meaningless because the subject could fail. Look at the significant implications of a successful result (please reread the outline for that).

 

 

[thesurfingalien]

It was not my intention to be offensive. If what I wrote came across that way I apologize!

 

The point is that in an ABX-test one does not (actively) have (as in an obligation) to listen to A. In all honesty, when differences are really big, I do not bother to listen to A. In effect, that comes down to the same thing you are suggesting. If B = X: no difference, if B X: difference.

 

By meaningless (probably a poor choice of word) I mean that it offers too much of an "escape" for people. If they fail the test, they are able to simply tell others the system was not revealing enough, the test-method was flawed because "fill in the rest" or the conditions did not allow them to repeat their performance.

 

I know this might sound paranoid, but before you spend a lot of time devising a test, I think this is something you should be aware of.

 

Please don't think I do no appreciate your efforts!

 

Regards,

Peter

 

 

 

[goldsdad]

No problem, I never saw your remarks as offensive. The "hmmmm..." was because at first I struggled to work out how I might have misrepresented my test idea.

 

I don't have the knowledge or training to devise a rigorous test, so there isn't much danger of me spending a significant amount of time on it. I'm enjoying the discussion in this thread and always appreciate feedback and corrections to my comments, anyway.

 

Thanks for showing an interest!

 

 

[Quest]

I skimmed through all the threads just because of so much talk on blind testing.

 

Well I didn't do this on AIFF vs WAV where I felt I wouldn't convert even if it did sound better just because of lack of metadata, but did it on my gf for AIFF vs ALAC and FLAC.

 

Test was done with same file, converted to ALAC/FLAC, or vice versa, and only 2 tracks were re-ripped just to try to eliminate variables. I only prepared 10 files. I went through each track, by playing back the 3 files randomly in sequence.

 

6 out of 6 times the gf pointed AIFF sounded better than the other two. After her first 2 tracks, she pretty much pinpoint which one she liked within 5-10 seconds into the track (having to back/forward and playback slightly longer on some to confirm), stating tone is the main difference. After the 6th one, she just felt irritated enough with me not to push my luck. I do this sort of testing with her regularly enough.

 

Why bother doing this AB testing with her? Differences are decent enough that I can hear it but since it was tone related, I'm not as good as she is as she's musically trained. As you know, people's preference in tone can swing either way, and I tend to use friends with music or recording experience to help me on my way.

 

When a musician tells me my music sounds 'real', that will be the day.. hasn't happened yet (they have commented 'real' on other systems when I take them around) though quite a few 'audiophiles' seem impressed. My dad seems to think its just because I have nice looking speakers and that creates a sufficient enough placebo to help some guys think the sound is great. I don't think I'm there yet myself. I have had only 3 visitors to my place since I installed the new speakers and all 3 want to buy the same one (2 already have).

 

Having said all that, I do hear some difference on AIFF vs WAV but it is less than AIFF vs ALAC.

 

Can we live without metadata? I personally worry about losing my database or having to reformat harddisk.. it's just too much work. As for AIFF vs ALAC, I don't see why we need to take the risk of lower fidelity since harddisk space is so cheap now.

 

[thesurfingalien]

Done :-)

 

 

 

[Sik_Lescinovid]

This is interesting. What software did you use to playback the files? How did she listen to them i.e. what was the audio setup? Did I understand correctly that you had prepared 10 different tracks each one present in 3 different formats (AIFF, FLAC, ALAC) and on each track you have randomized the order in which you played the formats? Was she aware of the randomization? And did I understand correctly that you interrupted the test after 6 tracks (because of her getting annoyed) with the result that up to this point she always chose the AIFF version as the best sounding? And finally was the test completely blind i.e. she had no chance of determining non-aurally what was being played?

If the answer to the above "yes/no"-questions is always "yes" then this mildly interesting: it's unfortunate that you couldn't go as far as at least 20 trials, only doing 6 trials doesn't yield any statistically meaningful results. But the fact that the test was (hopefully) done blind is a good start.

 

[thesurfingalien]

Hi Forrest,

 

I hope you can help me once more (I know this whole subject is not your favorite).

 

Can you explain what you mean by micro dynamic shading? I have no idea what that should sound like.

 

Thanks,

Peter

 

[4est]

Hello Peter,

 

By micro dynamic shading I mean that it can resolve subtle dynamic changes. If you are familiar with the term PRaT, it would be a sub set within that. You will know it when you hear it, but it is both more lively and sublime concurrently. Less homogenized/uniform. If one is ever so slightly off tempo it is detectable. Many dacs and jitter itself tend to mask this.

 

[wdw]

When I first saw some of the WAV vs. AIFF threads I thought it was a record of a certain kind of audio-nutbar madness. Due to some odd circumstances I am reloading my Library into a new headless mid-2010 Mac Mini and one nagging thought was on this issue. My default was custom AIFF, with error correction, to 16 bit, 44.1.

Today, to satisfy my curiosity, I took a great jazz album and wrote it into iTunes as both an AIFF and WAV. Happily the songs nested into each other, AIFF then WAV, the whole album repeating. I have a pair of Etymotic ET4's with custom cast ear pieces (highly accurate thou bass shy) and listened to this album on an iPhone 4 while riding my bike into the office. Then, needing to have some level of hearing beyond these totally noise isolating ear inserts, used the B&W P5 headphones, in my office.

WAV is clearly superior. Sad to say this because album art is so great to have. Looking forward to hearing these conclusions through the main system...btw..when is Berkeley going to finally introduce the Alpha DAC?

Thanks all of you on this board for the sharing of great advice.

WDW

 

[GB]

Colleagues, if you have the same text file in, say, zip, rar and 7zip archives, then uzip-unrar-un7zip archives and get different texts - your comp is seriously ill... But if you extract these archives "on the fly" to read this text from the stage - Hamlet's monologue, for instance - with specified speed (such as playing flac vs alac vs ape) and your comp is slow enough not to be in time - some sillables in the text may be missed with some archive format, if compression is high or THIS decompressor isn't optimized for THIS operational environment . BTW, it can cause the difference of players - WinRAR, maybe, extract ZIP files slower than WinZip, need more CPU power and so on... So, if your comp isn't faulty or slow - you must not hear any difference, playing music in various file formats or with various bit-perfect (!) players. IMO.

And if you hear the difference - heal your system...

 

[4est]

What is your point really? Have you tried it or is this just an intellectual exercise for you?

 

"So, if your comp isn't faulty or slow - you must not hear any difference, playing music in various file formats or with various bit-perfect (!) players. IMO.

And if you hear the difference - heal your system..."

 

WTF, do you think we are all daft? The whole point of this thread is that there are some of us that distinctly hear (the same?) differences, even though intellectually we know there should be none.

 

Can't and won't are different, save the "must" for your kids.

 

[GB]

English isn't my native language. Maybe I've used "must" in a wrong way. Sorry, 4est, I beg your pardon.

I'm in audio since 1968 and in IT since 1978. May I say a couple of words here?

Thank you.

 

[Paul R]

First, music is not a text file. A text file is not sensitive to timing or anything but gross transmission errors, while music appears to be very sensitive to even the tiniest of timing errors.

 

Second, even the slowest of modern day computers can keep up with playing even high resolution music files. The processing speed in and of itself is not an issue. The entire processing signature of a system may, and probably is a factor.

 

Third, with computers that are neither faulty or slow, people do hear differences.

 

You would probably be better served to start from the differences, using their existence as an axiom, rather than trying to prove that the differences do not exist.

 

Indeed, the evidence that people hear clear differences is quite compelling. That is not to say one format or the other is better, nor to draw any other kind of value judgement. Merely that the differences do exist.

 

If you can produce a new theory that explains why the differences exist, then I am quite certain many people here will volunteer to test the theory for you, within reason.

 

-Paul