AIFF Vs. WAV

[kennyb123]

I have no need to understand why these rippers produce files that sound different, so I really can't offer an explanation. I gave up trying to explain this insane hobby a long time ago. :-)

 

The differences between these rippers are clearly audible on my system unfortunately. I would have much preferred not coming to the realization that I needed to re-rip everything. My findings were confirmed by a bunch of friends who have great ears and highly-resolving systems. They all ended up switching to dbPoweramp.

 

 

[Paul R]

After all, nobody really needs to prove anything to anyone else about their personal systems, so long as they enjoy them.

 

I am one of those people where the itch to understand will nag at me incessantly though, at least until I take a good scratch at it.

 

For example, lately I have been eating up free time studying to get proficient enough to calculate and simulate a signal in a speaker cable.

 

That rabbit hole gets deep - veerrrryyy quickly!

 

Amazing what I am learning about speaker cables though. Cable sound differences are definitely NOT a bunch of hoodoo. At least not all of them.

 

Neither, I suspect, are the differences you hear from rippers. When I say I am unconvinced, it isn't anything personal. It just means that the rational part of me, the part that tries to keep me out of scams and expensive mistakes, is skeptical. The music-loving emotional part of me is totally open.

 

-Paul

 

 

[4est]

I have not dared enter that realm as I fear having to re rip everything. In my case it was not a ripper that caused the difference that I heard. I used plain old iTunes, but as a check I compared the original AIFF to the AIFF that I made from the WAV (that I made from the original AIFF). We couldn't hear a difference then. I am not saying that there are not differences in ripper outputs, just that I doubt that is what I heard.

 

[jhwalker]

I'm an IT guy, and I know about computer files - they either are or are not identical, and it doesn't matter which product originally produced them. So if it is shown that a rip from XLD vs. dbPowerAmp (or whatever) are, indeed, identical, they are actually the *same* file, and cannot sound "different".

 

It's not like the analog world of copiers / scanners or dubs from LPs / tapes, etc., where different copies / scans from different sources can look very similar but still be different - if the digital content of a file is the same, it is the *same*, period. And if the files are *not* completely identical, it won't be a different "feel" or sound throughout the whole file - the files will sound exactly the same except for the portion that is different. There is no room in digital for "tighter bass" or more "airy" treble, at least due to being ripped by two different rippers.

 

I'm not averse to saying that two files produced by different products *could* sound different if they are not, indeed, identical (i.e., if there is a bug somewhere along the way that causes the data to be completely wrong *throughout* the file), but I'm the kind of guy who trusts bits / bytes more than my eyes / ears. If my eyes or ears tell me something I know cannot be, I know my eyes and ears have been fooled.

 

So while I know some rippers are better at error-handling (or at least letting me know if they find an error) and some lossless file formats take up more or less space, I'll remain agnostic about any purported differences between them and use whichever one is most convenient for my purposes

 

 

[4est]

There is still a long chain of custody in a file from the hard drive to the pins on the DAC chip. To me, my test indicated that it is not the file checksum changing, but how the file is handled/interacts with the device(s).

 

"but I'm the kind of guy who trusts bits / bytes more than my eyes / ears. If my eyes or ears tell me something I know cannot be, I know my eyes and ears have been fooled."

 

At first I felt that way about this. Age has taught me to respect my senses and gut, even when I do not understand. Advancement requires an open mind. No offense intended...

 

[blueixus]

"I'm an IT guy, and I know about computer files - they either are or are not identical"

 

To echo similar comments life appears not to be this simple, whilst they may be 'the same' by IT standards the computer appears to be reacting to them in different ways.

 

Computer Audio is at a very early stage, there is a lot more to understand as to why and how things sound better/different.

 

At recent HiFi show I was talking to one of the driving forces behind one of the popular playback programmes.

 

They told me it is only recently they have begun to understand why software that produces bitperfect output can sound different for others that also produce bitperfect output.

 

 

 

 

[goldsdad]

@4est

 

But your player, Audirvana, in "memory" mode completely removes any element of, as you said, "how the file is handled/interacts with the device(s).".

 

Here is my understanding of how Audirvana in "memory" mode works.

 

When Audirvana plays a file, either a WAV, AIFF, FLAC or ALAC, before actually starting to "play the music", it extracts all audio data from the file and decodes it to a sequence of raw sample values, storing the values for the entire track in a buffer in memory. Remember, the series of values in that buffer is absolutely independent of the format of the original file (or the existence of metadata or artwork in the file). Exactly the same series of numbers will result from deconstructing WAV, AIFF, FLAC or ALAC file equivalents of a track.

 

This is in contrast to, say, Decibel, which does read an entire file into memory, but only decodes the file piecemeal while it "plays the music".

 

At the moment that Audirvana actually begins "playing the music", the original file effectively has ceased to exist, and that file will not be processed further. Only the buffer containing a series of numbers, not a file structure of numbers, is played now. These numbers have no memory of where they came from, and look exactly the same whether they were once-upon-a-time stored in a WAV, AIFF, FLAC, ALAC or any other kind of file.

 

 

 

 

 

 

[dchan88]

From reading all the posts, I found those who said there is no difference between FLAC and WAV are defending their position more passionately and some time insulting those who can hear the difference.

 

Generally in sound system, if you cannot hear what other people said, it does not mean it does not exist. Every sound system has certain level of resoluion or transparency. There is no point to beleive your own system is the best and nobody else can have a better system.

 

In my system, WAV sounds better than FLAC.

 

[4est]

I realize that, hence my statement about hearing it with Audirvana too. Obviously it doesn't completely do it, or put another way, stores the difference in memory. As I have stated, I do not know what is causing this. My observation about the ESS9018 chip is only because I do not know what the cause might be, but I know of at least two other individuals with the same DAC as I whom can hear it.

 

I am pleased that some others have chimed in on this in so that the OP might feel compelled to try it themselves. If I were more of a betting man I'd make a strong statement such as "If you do not hear it, look closer at your signal chain/system.". As it stands, it might just be my hardware. The only thing that I do know for sure is that this is not in my mind, but in my ears.

 

[Paul R]

But I suspect in the case under discussion, the two files are not identical.

 

Perhaps one is big endian the other little endian. Won't make any difference at all when it converted for play, but the actual conversion itself could cause an audible sonic difference. Such files would even pass a null test, in theory, but could reasonably sound different. Also, it would make sense for iTunes/XLD to store data in big endian format, and dbPowerAmp in little endian.

 

This is the same (rather, a similar) mechanism posited to make ALAC and AIFF or FLAC and WAV files sound different.

 

Not saying that is the case of course, but it is a theory that can be tested and proven wrong, which is how we go about learning things. It could easily be some other explanation, though what that "other" explanation might be, I have no idea whatsoever.

 

Remember, there is always a third path, where both the hard facts of digital data and your ears are both right. We just don't know enough yet.

 

-Paul

 

 

 

[kennyb123]

Paul you are absolutely correct - nobody needs to prove anything to anyone else about their personal systems. My goal in raising the point about rippers is only to save folks from having to re-rip. I wish someone would have brought this to my attention before I had embarked on the effort to rip all my CDs using iTunes.

 

Also, while I can understand the emphasis on basing our opinions on what we hear (or don't hear) in our systems today, I think we should keep in mind that our systems will likely get better over time. I didn't hear a difference between ALAC and AIFF rips using dbPoweramp. But I decided to go with AIFF because I felt there was a chance that a system upgrade might eventually allow me to hear the difference. Likewise, if someone doesn't hear a difference from rippers today, little harm will come from remaining open to the idea that a system upgrade may make these differences quite obvious.

 

The other reason I brought up the rippers: the improvement from dbPoweramp helped to improve my enjoyment of digital. A glassy edge was removed leaving a more natural sound. I was thrilled by this as the improvement cost me very little - except time and the dbPoweramp license. It's rare in this hobby to come by upgrades like this. For that reason I think it's worth spreading the good news.

 

 

 

 

[jhwalker]

@goldsdad

 

Exactly - once the file is decoded, it is the exact same series of 1s and 0s - this can be proven by comparing the two files. At that point, it *cannot* sound different depending on where the 1s and 0s originally came from - it just doesn't work that way.

 

I have no problem positing a chain of software / hardware that reads the files differently (or handles the file headers inconsistently or doesn't respond properly in a big-endian / little-endian situation, etc.), but there is no *intrinsic* advantage between WAV, AIFF, FLAC, ALAC, etc. - they are just various containers / compression schemes for identical data.

 

If *your* software / equipment chain reads WAVs "better" than FLAC, that's one thing and I certainly wouldn't dispute it . . . but that has nothing to do with the value of WAV vs. FLAC in the larger scheme of things.

 

[goldsdad]

@Paul

 

Audirvana is a special case amongst audio players, and can help solve these riddles, in my opinion.

 

In the case of Audirvana in memory mode, it should absolutely be of no consequence whether files are big-endian or little-endian, or uncompressed or losslessly compressed, or contain metadata and artwork or purely audio samples. The files are not being played. All decoding into an absolutely file-format-independent series of numbers - the track buffer of raw audio samples - is done before playback begins. Only after that stage does playback of the raw audio samples, and not a file on disk or a file structure in memory, begin.

 

Perhaps there's a bug in Audirvana, such that not all files are correctly being decoded into the track raw sample buffer before start of playback. That could result in a buffer that's different for a track that was extracted from one file versus another file when the two series of numbers in the buffer should be identical.

 

The audio community needs to discover if ears truly are right when they hear something which currently seems illogical. Those who insist that their ears are infallible and can detect differences that are currently inexplicable when using Audirvana memory mode could step forward and be tested. If these people are found correct beyond reasonable doubt, then explanations can be sought, hopefully advancing factual knowledge and improving audio for everyone.

 

 

 

[dchan88]

I don't think such a challenge is neccessary. It is possible that the system you setup for such a comparison is not transparent enough to resolve the difference.

 

[goldsdad]

Test the people claiming to have infallible hearing with their own systems in their own homes using music with which they are intimately familiar. It wouldn't be difficult to set up a randomised playlist of multiple pointers to two files which are basically two different containers for one common set of audio samples.

 

 

[4est]

Goldsdad:

Why on earth would it be up to me to prove something to you about what might help YOU? As for explanations being sought, they already are by people far more knowledgeable than I (and possibly you).

 

I do not know where you have been, but HP and others pioneered the notion 40yrs ago that what can be heard cannot always be measured. FWIW, I have never claimed that my ears are infallible. I only claimed that I (and three others) have readily discerned differences here, and WAV always wins. It isn't even subtle if you are accustomed to the system as I can tell from the adjacent room. Oddly enough, my second round of tests was done completely on Audirvana at Damien's explicit request. I realize that you do not want or cannot believe this, but alas...

 

This isn't IT, and we are not in Kansas anymore.

 

[Paul R]

Yep, that is a good way to think about it in my opinion.

 

I don't use Audirvana because I am really dependent upon good remote control, and the REMOTE app on our iPhones and iPads is the best remote I know of right now. Hence, Amarra or Pure Music is our choice.

 

Wouldn't using Amarra with Cache or Pure Music with Memory Play (not Hybrid memory play) be the same as using Audirvana?

 

-Paul

 

 

[Sik_Lescinovid]

Here's what I did a while back (and I recommend anyone considering reripping his whole library with a new ripper do that too): take 20 to 50 albums (depending on how much time you have, I took 28 back then) and rip them all to FLAC (or WAV if you're impatient) using as many different rippers as you please (but all using test-and-copy and AccurateRip if available) and make sure all the rips are "perfect" according to the hashes of the respective rippers. Now convert all the FLAC files to WAV using a single converter (or you can use the same ripping software again if applicable, only mix it up as to avoid converting with the one you ripped, make sure you mixed it randomly). Now you'll have about a thousand or so songs in WAV, each one present in as many versions as you have used different rippers (you could do the whole thing with one song but the point is to strengthen the statistic by using helluva lot of songs). Next you should transfer all the files to Linux (if you're not there already), what I did back then was rip using XLD on a iMac, dbPoweramp on my Windows laptop with a crappy drive and cdparanoia on Ubuntu. Anyways, once you have the files on Linux (although I'm sure it can be done on other operating systems as well) run a bit-for-bit comparison on all the files such that each song gets compared to all of its other versions. Before I give you the result of the comparisons I want to say that the reason I suggest using Linux is because such batch processing jobs (including a lot of batch renaming and folder generation etc.) can be done quite easily on an operating systems equipped with an analog of a Linux bash shell. Also you will note that the procedure was designed to address ripping and conversion with different programs at the same time. If we have differences at the end it's unfortunately hard to track down the culprit (because of so many stages involved), but if there aren't any differences the result is so much stronger since it eliminates a whole range of supposed sonic mutilators at once. Anyways, the result was that all versions of each song were bit-for-bit exactly the same. I must admit I was a wee bit surprised back then as I expected at least the OS differences to introduce some minor bit-differences in the headers or so, but it was not the case. What can one conclude from this? Well, all the converters and rippers used generate EXACTLY the same output. Doesn't even matter if I convert FLAC to WAV with one or the other. The gist of the story: whoever claims he hears a difference due to different rippers is (to anyone who has done the above) either under a very grave misapprehension of his own abilities to properly set up software or imagining things. Since I've devised the above procedure and went trough it I will say the following to anyone in this thread claiming that it all depends on the hi-fi system or that I couldn't possibly hear a difference because whatever sorry excuse: Either you should recheck all the setting and make sure you haven't accidentally compared WAV files and 128kbps MP3's or you're simply imagining things. And I can say that with full conviction.

 

Finally note that the bit-for-bit comparison I ran on Ubuntu is (unlike hashes) completely deterministic and not probabilistic in any way! I.e. if the comparison says two files are the same then they are the same with absolute certainty (unless there is an invisible god with supernatural powers who is playing a very cruel prank on you, which is a case I simply ignore due to irrelevance).

 

[kennyb123]

goldsdad wrote: "The audio community needs to discover if ears truly are right when they hear something which currently seems illogical."

 

Ever since the first audiophile DAC hit the market, there have been folks telling the audiophile community that they can't be hearing what they claim to be hearing. CD was said to be "perfect sound forever". And over and over again we were told that 16/44.1 is sufficient - and that higher resolutions weren't going to offer better sound.

 

A lot has changed since then. Digital has improved dramatically, proving the naysayers very, very wrong. What we should learn from this is that there is a lot more going on in digital reproduction than meets the eye. What seems illogical today might very likely be considered perfectly sound reasoning at some point in the future.

 

 

 

[goldsdad]

"I have never claimed that my ears are infallible [...] I can tell from the adjacent room."

 

Yeah, I know that's twisting context, but it's only meant as a joke.

 

I can see this thread getting locked now.

 

 

 

 

[4est]

Yes, I am done here now. I have caused enough ruckus for the day...

 

[goldsdad]

@ Paul

 

"Wouldn't using Amarra with Cache or Pure Music with Memory Play (not Hybrid memory play) be the same as using Audirvana?"

 

That would be the same if, and only if, these apps completely decode an entire track into its raw samples before commencing play of these samples. That's a distinctly different process from simply reading the file from disk into memory before commencing play, then decoding each small chunk of the in-memory representation of the file while simultaneously playing the previous small chunk.

 

I'm sure Decibel takes the chunk-by-chunk approach, therefore I am happy to concede that file format could have an effect on sound with Decibel. I have never investigated Amarra or Pure Music, so can't say what they do.

 

 

 

 

[Paul R]

Well, he holds a lot of opinions that are controversial, and many folks, including me disagree, with many of them. HP has an agenda all his own.

 

If there is a real difference, it can be measured. The mystical stuff is just smoke and mirrors, usually used to defend an position that suffers from a lack of basic information. Often, the subject suffers that lack of information because tools to measure the difference are not available, or sometimes, we don't even know what it is that has to be measured! Or it's just too darn expensive to do the measurements.

 

The flip side to that, is people like Forrest and Barrow, among others, can and do reliably hear differences other people miss. I doubt it has anywhere near as much to do with their systems as they credit, and far more with the way they have trained themselves to hear. There is no mystery in a person training themselves to discriminate sounds better than average.

 

It would be foolish to discount what they, or anyone else, claims they can hear, even if we have not figured out how to measure it, or are unwilling to spend the dollars to measure it. And it forces those folks into a corner where they feel unjustly put upon.

 

Not to mention it is a bit rude to do so. I joke gently about it sometimes, because a lot of the WAV vs. AIFF stuff is a Microsoft vs. Apple thing, whatever that is really all about.

 

-Paul

 

 

[goldsdad]

"[...] people like Forrest and Barrow, among others, can and do reliably hear differences other people miss. I doubt it has anywhere near as much to do with their systems as they credit, and far more with the way they have trained themselves to hear. There is no mystery in a person training themselves to discriminate sounds better than average."

 

Yes, that's absolutely right. With all due respect to Forrest and others, anyone and everyone can and does make mistakes at times, regardless of expertise.

 

Of course, nobody has to prove anything to me. The point of formal testing, is that all audiophiles, including Forrest, could benefit if those claiming exceptional aural discrimination of file formats were to give developers an indisputable avenue of research on which to focus.

 

 

[4est]

Jeeze, here we go again.

 

I would relish being proven wrong or find the problem with my rig if that is the case. I would much rather replace equipment than re rip or re work things for WAVs or what have you. The problem is that non of the BS that is coming forth offers one bit of evidence to make me question my findings. As for formal testing, I live in GR Michigan, c'mon over, the water is nice this time of the year.