View Full Version : AAC and WAV - Taste the difference.
Can you tell the difference between an AAC lossy file at 256kb/s and its parent WAV file?
There is a test running on Zerogain at the moment. Marco has suggested I invite you to:
Sign up.
Download.
Listen.
Vote.
Wait to hear if you got it right.
Look here:
http://www.zerogain.com/forum/showthread.php?t=25890
I've just compared the first few seconds of the Monheit tracks so far, with headphones off the laptop. Pretty clear differences to me, the file that I think is the wav sounds more open and cleaner and with more natural sibilants.
Stratmangler
08-08-2010, 23:03
Can't see the point of comparing lossy with non-lossy files, as the lossy files will always come out worst.
AAC files sound better than MP3 files of the same bitrate.
A WAV file of the same track will always sound better than any lossy compression method devised to date.
Pointless exercise IMO.
Stratmangler
08-08-2010, 23:22
Can't see the point of comparing lossy with non-lossy files, as the lossy files will always come out worst.
Why AAC ?
Has this got something to do with iTunes shop ?
Why on earth don't iTunes do the decent thing and admit that FLAC is a valid format ?
More to the point, why don't they support it ?
Can't see the point of comparing lossy with non-lossy files, as the lossy files will always come out worst.
AAC files sound better than MP3 files of the same bitrate.
A WAV file of the same track will always sound better than any lossy compression method devised to date.
Pointless exercise IMO.
At what bitrate do the differences become vanishingly small?
It is not a pointless exercise, unless you are prepared to maintain your position in the face of data collected in a properly staged blind subjective test.
Why not listen and vote?
Rob Holt's first test on Zerogain produced a 50/50 split in a sample of 20.
This test is more comprehensive and it is hoped the sample will be much bigger. The first test used MP3, as you say AAC does better at a given bitrate, so the bar has been raised.
I don't imagine for one minute that Apple has any involvement in this test.
I agree with you, Apple ought to have the grace to allow iTunes to play other formats, even if that impairs Apples total integration policy. Let the user decide.
But let's not get sidetracked.
Please listen and vote. In the first round some who were absolutely convinced they could always tell the difference got it wrong.
Whilst this is a blind test it's not thoroughly useful or repeatable. It would need to be staged on one system or a specified collection of systems. Here the sort of thing I'm getting at:
Using Beresford 7520 DAC via USB I couldn't detect the difference between PC to DAC connection via 10m of USB cable with repeater or a 1m USB cable. Having done the "passive mod" you'd have to be truly deaf not to hear the difference.
So what I'm saying is, how useful and reliable can the results of this AAC vs WAV poll be? Any particular vote simply says what that individual heard via their system. Some systems may not be good enough to allow the listener to discern a difference, some listeners may only want background music and don't hear the subtlety WAVs may bring over AAC. The poll is a little interesting but it can't give any really meaningful insight, there are too many variables.
Whilst this is a blind test it's not thoroughly useful or repeatable.
Those point's I have already made on Zerogain. The test will "prove" nothing, but it could be fun; and it could demonstrate to some individuals that the matter is not as clear as they would like to think.
Interestingly, the folk at Harbeth have said that they cannot easily tell the difference between a 128kb/s MP3 and a lossless file.
StanleyB
09-08-2010, 08:53
Obviously owners of a TC-7520 or Caiman with the passive mod have an advantage over the majority of the listening public when it comes to hearing differences. I can pick up the difference between a FLAC and a WAV file now.
Try re-constituting an MP3 file back to red-book and see what happens... I don't think it's as cut and dried as these tests suggest.
Try re-constituting an MP3 file back to red-book and see what happens... I don't think it's as cut and dried as these tests suggest.
What does that mean, Dave?
Oh dear, and we were all doing so well; no ridiculous claims and sensible comments :donk:
StanleyB.
“I can pick up the difference between a FLAC and a WAV file now.” :brickwall:
My understanding has always been that FLAC is lossless. That means, for those that don’t understand, there is no audible difference before and after compression ;) Thousands of very serious audiophiles rip their reference music collection to FLAC. Some may use WAV as their current play rips (many computer audiophiles rip two copies, one for reference and one for play in case the play files get corrupted or a hard drive fails) but I have never heard anyone state they could hear a difference between WAV and FLAC.
If you can indeed tell the difference StanleyB then may I suggest you warn all those audiophiles they’re doing it wrong.
dave2010
09-08-2010, 10:25
John
I don't know whether Stan can really tell the difference between FLAC and WAv, but I wouldn't assume he's telling porkies. If he can, it's probably something to do with the software used to play back the files, since, as you have noted, the actual data should be the same. There must be more processing to pay back FLAC files from a data stream, and this could, perhaps, make a difference to the sample timings.
What would perhaps be of interest is if he could tell us whether he could tell the difference between a WAV source, and a WAV file which has been "round tripped" to FLAC and back. Since they should be the same, that'd tell us something.
I agree that if a difference is heard with FLAC then it most likely is due to an artefact of the processing required to play it. One of the factors with PC audio is that the sound is very sensitive to set up and the hardware used.
Dave2010 and Clive.
In the cases you outlined it’s not the difference between Flac and Wav then is it? I’ts the difference between players or other components.
Dave2010 and Clive.
In the cases you outlined it’s not the difference between Flac and Wav then is it? I’ts the difference between players or other components.
Well, yes and no. We're really splitting hairs here....you're right in that I have no reason to think that FLAC and WAV should intrinsically sound any different. But you could also argue that FLAC may place an extra load on the hardware which could theoretically result in some sonic differences, presumably on a low powered PC or netbook. It would not be FLAC itself that sounds different but the instead the consequence of using FLAC.
This is all theoretical and very possibly of little consequence in most cases.
Fair point Clive ;)
I have £50 here for anyone that can prove they can tell the difference in a controlled test ;)
Between FLAC and WAV or between AAC256 and WAV? Or both?
From what I've read (and Brian should be able to confirm this), it seems to be to so with the software rather than the files themselves.
In digital, bits is bits as long as they're not corrupted. I believe it's how they're read and clocked out of one box and into another perhaps?
StanleyB
09-08-2010, 13:16
Oh dear, and we were all doing so well; no ridiculous claims and sensible comments :donk:
StanleyB.
“I can pick up the difference between a FLAC and a WAV file now.” :brickwall:
My understanding has always been that FLAC is lossless. That means, for those that don’t understand, there is no audible difference before and after compression ;)
Aha, you are implying that I am a liar :eek:.
Well, I challenge your assumption that FLAC is lossless, and substitute my own: There is no such thing as a lossless analogue to digital audio conversion process. So there is no way that a FLAC file can be lossless.
Worse than that: you accept blindly that FLAC is lossless and that there is no sonic difference between the original (which original by the way, if I may be so bold in asking?) and the FLAC end result. However you do more than question my statement that I can hear a difference:mental:. Has the inventor of FLAC ever stated that there is no sonc difference between FLAC and whatever the original audio signal was that he/she copied? Or have you been repeating verbatim what you have read here and there? I do not recollect the inventor ever making such a statement, but I am open to any link you may wish to provide where he/she makes that claim.
OK, I accept that thousands of people have been using FLAC and have been repeating the claim that it is lossless. But may I draw your attention to the millions of claims that Father Christmas exists, as do elves and the stork that delivers the babies:eyebrows:? Tell me that those millions of people are also right when I say they are telling fibs.
I regularly read the excuse that it is the software that is at fault if FLAC sounds any different to whatever it is copied from. Why blame the software? Why not the hardware, or the operator, or the suitability of the original source material? Is it because it's far easier to finger the software, or is it because most people would nod their head in agreement without resorting to any further investigation to confirm any such claim?
Maybe you should invest in a Caiman, John, and experience for yourself how easy it is to differentiate FLAC from WAV, etc. When the right tool is now available and at an affordable price, there is no more need to 2nd guess ;).
This reminds me of a well known member on AoS who questioned whether the TC-7520 would indeed improve the quality of his listening pleasure. As he crawled under the table to plug in the TC-7520 into his Freeview box, he shouted to his son @ I can hear the improvements from under the table.... Not so long after that I modded his unit, which he opened up and tried to figure out what I had done. On failing to see anything obvious, he questioned me about whether I had done anything, especially since he had paid for the mod. I refunded him the payment and stopped doing any mods for anyone since then. As time went by, the results of the mod did start to filter through, but the damage to my credibility had already been done.
But as the countless numbers of Caiman owners have found out: every claim I have made about the Caiman has been received with scepticism first, and utter amazement to their actual correctness afterwards :). So guys, dig out your FLAC files and let the Caiman with passive mod show you the correct way to nirvana :cool:.
If you decide to expand a FLAC file that was derived from a WAV or AIFF file, you will get back the exact same data as the original file, hence the definition lossless. That is a fact. Same with Apple Lossles, ALAC.
The assumption, however, is that because the FLAC/ALAC file contains the same exact data, therefore it *must* sound the same.
Does it? And if it doesn't, where is the flaw?
Sad to say but bit perfect is only the beginning of truly good audio............
CD
:scratch: Sorry Stan, but I don't believe that anyone can hear the difference between a WAV file (for example ripped from a cd) and a FLAC file converted from the WAV file (original). FLAC indeed is a lossless format as well as WAV; the recording devices (mics) however are not lossless.
To prove that one can hear the difference between a FLAC and a WAV file, one should compare the original WAV file to a WAV file that has been converted from a FLAC file converted from the original WAV file. I believe that those too files are identical as digital music is only bits as mentioned before, and if the bits are exactly the same in both files, there is no difference between the files' audio content.
Could you do this kind of ABX test to prove that there is a real difference that hasn't been caused by the extra processing power needed, software or by a placebo effect? I have never heard of anyone who could hear the difference.
Reid Malenfant
09-08-2010, 14:40
I think what Stan is trying to say is that digital audio isn't lossless fullstop ;)
By definition an analogue signal has an infinite number of positions on a waveform whether that be in time or voltage. As soon as it has been converted to digital there are no longer an infinite number of voltages positions available, but a finite number & so many samples per second rather than an infinite number :)
Therefore digital audio is lossy, some forms to a lesser degree but in the end it all is :eyebrows:
Therefore digital audio is lossy, some forms to a lesser degree but in the end it all is :eyebrows:
Of course, and an analogue transmission system will degrade the music too, but in different ways. It therefore follows that an analogue transmission system is also "lossy".
But we are stretching the meaning of "lossy" and "lossless" which is quite clear and given by context.
That the digital compression and decompression employed in FLAC, ALAC and other "lossless" formats does indeed return the data to its original state ought to be a matter that can be objectively measured. I thought it had been proven that a FLAC file "unzipped" is identical to the file that was stuffed into the "zip". The process is not in principle different to zipping and unzipping other computer files.
I cannot make a judgement about what Stan is hearing.
Reid Malenfant.
There is no “I think what Stan is trying to say” about it; hence my direct quote.
I don’t think anyone would argue that sampling a wave form incurs a loss of data. The salient point is whether or not that lost data makes an audible difference.
Dealing with the sampling of a waveform first;
http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem
The term “lossless” could perhaps be misleading for the exceptionally pedantic. However, most seem to realize that this term refers to what can be heard (quality), not to a data quantity. Of course there will always be those who say any sampling process, no matter what frequency the sample will give inferior sound quality, I recall being in that particular camp myself many years ago.
What often seems to be forgotten is that the majority of Dacs also either over sample or up sample and that we have been waxing lyrical about expensive CD player’s sonic qualities for many years now happily listening to 16/44.1.
That’s great StanleyB. They do say faith is half the battle ;)
However, no matter how hard I try to convince myself I can fly when I stand on that cliff edge with all the conviction I can muster science always rears it’s ugly head and says to me “John, don’t do it. What will happen is no matter how hard you flap your arms you will accelerate downwards at 9.81 metres per second per second until you reach your terminal velocity and spend the rest of the journey screaming your way to a messy and ugly death.”
“After you StanleyB my good man” is all I can say to your doubts concerning scientific theory and a considerable amount of opinion gained through knowledge of the outcome of people threw themselves of cliffs.
(Given the tone of Stanley’s post I hope I will be forgiven for a somewhat blunt response)
With regard to your arrogant and blatant product advertising and your suggestion that I should buy one of your Dacs to test your assertion for myself I have to wonder why on earth I would want to replace the very competent Dac I have now with an inferior product?
StanleyB
09-08-2010, 16:23
Reid Malenfant.
There is no “I think what Stan is trying to say” about it; hence my direct quote.
I don’t think anyone would argue that sampling a wave form incurs a loss of data. The salient point is whether or not that lost data makes an audible difference.
Why the sudden back tracking after banging your head against the wall :scratch:? I also noticed that you are now questioning if that loss of data makes an audible difference. Anyone else in stitches over their keyboard :lol:?
I think it's time to stop, Stan.
StanleyB
09-08-2010, 16:39
To stop what exactly Brian? Questioning the integrity of someone who implies I am a liar? The cock has not crowed three times yet...
StanlyB.
To be frank, I don’t think you have managed to follow the qualification and reasoning in my posts. There has been no “backtracking”. If you find my style of writing and reasoning hard to follow then I can only apologies and attempt to keep it simpler.
The point in question is whether or not you or anyone else can tell the difference between a Flac file and a Wav file when played through the same medium.
You say you can.
Not only do you say you can but you say you can primarily because of the qualities of the Dac you make.
However, it seems the vast majority of audiophiles can’t.
In fact, you are the only person I have yet to come across that states they can hear such a difference.
Attempting to gather forum support through being an established member having made a bit of a fool of oneself is a very old and generally despised tactic.
If you are unable to stand up for yourself StanlyB using polite/humorous/intelligent/informed points of discussion then I suggest you pass this debate by.
Vincent Kars
09-08-2010, 18:26
I’m afraid I duplicate a bit what others has said before.
Take a WAV file
Convert it to FLAC
Convert the FLAC back to WAV
Load both in an audio editor, time align and subtract the 2 tracks.
You end up with a track with zero’s only.
Yes indeed, FLAC is lossless, just like WinZip or any other well implemented compression software.
A slightly different test:
- Record the SPDIF out in both cases
- Load both in an audio editor, time align and subtract the 2 tracks.
Again, you end up with a track with zero’s only.
B.t.w. this has been done recently to test differences between iTunes and Amarra (playing of course both recordings at the same and native sample rate).
This is probably the most convincing test as the SPDIF out is what goes into the DAC.
All these test yield the same result, both files are bit identical.
If 2 files are bit identical, can they sound different?
PC based audio is PCM audio, you have the samples and the sample rate.
Correct reproduction requires to things:
- the right sample, any alternation to the bits will alter the sound
- the right time, any unaltered sample arriving to early or to late will alter the sound
Yep, the second one is known as jitter.
A logical conclusion, as PCM audio consist of 2 components, both must be right.
Any comparison of possible differences e.g. between audio formats should take both aspects into account.
What happens if you play WAV?
A codec will interpreted the content, translating it to raw PCM.
As the content is almost identical to raw PCM, this codec has almost nothing to do so has very little impact on CPU use.
However, as the file is uncompressed, you do have a lot of I/O
What happens if you play FLAC?
Exactly the same, a codec will interpreted the content, translating it to raw PCM.
However, as the audio is compressed, it must be unpacked. This requires more CPU.
The file is about half size compared with WAV, so I/O is substantially reduced.
There are claims that FLAC sounds different and this difference is often qualified as worse than WAV.
Do we have to conclude that CPU activity is more damaging than I/O?
There are a lot of claims that I/O is detrimental to, one should use a memory player, loading the entire song in memory before playback starts.
What would happen if we translate the WAV and the FLAC to PCM first, load the result in memory and start playback. Will they sound different?
My personal opinion:
DA conversion is a delicate process.
Claims about the audible threshold of jitter varies but we probably talk nano or even pico seconds. http://thewelltemperedcomputer.com/KB/BitPerfectJitter.htm
I think it is perfectly possible that electrical activity going on inside a PC disturbs the clock of the sound card so maps itself into sample rate jitter.
Increased system activity will decrease the sound quality.
This is pretty much like having a video card and the more system activity, the more your screen starts to blur.
One might argue that if sound quality fluctuates with system load, this indicates a design flaw.
As a consequence, on a well designed system you won’t hear any difference and on the ones with a crappy sound card, you do.
Vincent Kars, :exactly:
Very well elaborated and clarified.
CD
Tony Moore
09-08-2010, 19:09
Seconded!
Vincent Kars
I must admit I find the whole subject quite fascinating.
I rather suspect that modern multi core processor hyper threading technology makes CPU loading largely irrelevant but I must admit I know of no studies that have thoroughly explored this issue.
On my desktop system I have attempted to measure CPU usage and RAM usage when playing both Flac and Wav files. The test apps are far from accurate but the usage of both RAM and CPU seems to be very light; not more than 7% of CPU and RAM hardly significant. As yet I have not detected any deterioration in sound quality even when running a system intensive program such as a video game without sound over a music player playing music files.
Interestingly, the use of an ASIO wrap (ASIO4all) in preference to a true ASIO driver or Kernel Streaming did increase CPU load by a small margin but still no audible deterioration in sound
Perhaps you are right and with a well designed system the issue is irrelevant.
I have managed to convince myself that music replay sounds better on my laptop which has all but the bare essentials stripped out and non essential services terminated but I wouldn’t like to be ABX tested on that even considering I seem to have a number of helicopters taking off inside the desktop case.
“As a consequence, on a well designed system you won’t hear any difference and on the ones with a crappy sound card, you do.”
Does the above mean you believe that a poorly designed Dac will give an audible difference between Flac and Wav?
Vincent Kars... I was just going to say that ;)
You know this argument of the computer having more than enough resources to handle digital audio always comes up. I do believe that a computer can handle computer audio flawlessly......if it was designed to. Fact is, the computer runs a pre-emptive operating system. So the OS lives to be interrupted. For audio, that's not good.
I think a good analogy for how a computer handles is this scenario:
Picture the incredible hulk bench pressing 10,000 lb weights. He has such massive strength that he can do it easily, gracefully even. Now, while he's doing the bench pressing, a cute little girl comes by and starts poking his ear playfully. "Aww, how cute and green your ear is" she says. The Hulk, trying to concentrate on gracefully pushing the massive weights up and down, continues to do so, but his gracefullness has sufferred by the irritating curiosity of the little girl.
I think that is the case with computers and audio. The constant interruptions add just a wee bit to the jitter in the audio stream.
Food for thought.
CD
Vincent Kars
09-08-2010, 21:26
I rather suspect that modern multi core processor hyper threading technology makes CPU loading largely irrelevant but I must admit I know of no studies that have thoroughly explored this issue.
I don’t either.
The funny thing is that on AudioAsylum, a lot of inmates go the CPlay way, a minimalistic low power system stripped to the bones.
Gordon Rankin (Wavelength) on the same forum is one who advocate you need a big machine. Indeed the multi core hyper threading
the usage of both RAM and CPU seems to be very light; not more than 7% of CPU and RAM hardly significant
Small wonder, a CD is about processing 600 Mb per hour. Most PC’s do consider this sleep mode!
Perhaps you are right and with a well designed system the issue is irrelevant.
I don’t hear a difference between WAV and FLAC.
If there is some back ground activity I don’t hear it either
Is this because:
- my gear is not revealing enough
- my ears are not discerning enough
- the recordings used lacks resolution
- the room acoustics are bad
- wrong power cord
- etc
No, it is of course because I have the right gear! Period!
To phrase it a slightly different, there are of course other factors than jitter only making us hear/not hear a difference.
I have managed to convince myself that music replay sounds better on my laptop which has all but the bare essentials stripped out and non essential services terminated but I wouldn’t like to be ABX tested on that
Twenty smilies, I do same knowing that it is probably nonsense but audio nervosa is hard to cure
Does the above mean you believe that a poorly designed Dac will give an audible difference between Flac and Wav?
I think you at least can’t rule out that RFI, EMI, ripples on the power rails , etc might disturb the DA conversion.
I know that products like the Ayre QB9 or the Bryston BPP-1 ( http://thewelltemperedcomputer.com/HW/PC.htm ) try to isolate the DAC as much as possible from the PC including galvanic isolation.
What we really need is the guy with that very expensive gear allowing to measure the jitter on the digital out. (I believe a second hand Wavecrest set you back $60.000, most boutique audio manufacturers can’t afford)
Then at least we have some hard evidence (or not ….)
Interesting thread. May I point out that Stan said this
I can pick up the difference between a FLAC and a WAV file now.
So can I. The difference is about 25% less space.
Seriously though, Stan didn't say he could hear the difference, he said he could pick up the difference, which may not be the same, or at least that's what he said first. I'd be interested to know in which way he detects a difference.
Codifus.
Interesting analogy ;)
Vincent Kars.
I also wouldn’t rule out the possibility that EMI and RFI could provide some distortion. I’m afraid I don’t have the estimated $60,000 either so it’s back to deduction, accepted empirical evidence, relevant scientific theory and collective opinion.
I’m also glad to note I’m not the only one who despite a reasonably rational outlook suffers from the occasional bout of audio nervosa :)
I think jitter, while a factor for any audio data transfer is less relevant with well implemented asynchronous transfer. Much will depend on the clock accuracy used by the client device and how effectively it communicates with the host clock.
I also believe galvanic isolation is more easily implemented using USB as the transfer medium and is extremely effective; hence my choice of current Dac
Despite my best efforts to ascertain exactly how HRT achieve their D+ and D- rail regeneration I have assumed it to be galvanic in one form or another.
Given programming asynchronous transfer protocol is beyond the ability of many Dac manufacturers, those unable, or unwilling to license from established sources are pretty much limited to SPDIF as their primary Dac to computer link.
I gave up on SPDIF quite early in my Dac search having tried out a few Dacs ( Beresford TC 7520 MF V-dac, various NOS Dacs, Cambridge Dac Magic and even an early Benchmark I borrowed) who use this as their primary means of connection. The most satisfactory results to my ears came from a Asus Xonar sound card strangely enough :scratch:
Unfortunately, I didn’t think to test the Flac v Wav using SPDIF with any of the above Dacs taking it as read that there would be no difference to be found.
When I get my Asus card back (Keith if your reading this!) I shall give a listening test a shot. I’m not very hopeful though because I can’t tell the difference between 24/96 and 24/192 either which I have been assured by some others is blatantly obvious if you have the right kit.
Audiophilia, it’s a strange affliction. I must remember to rebook for further therapy :)
I'd suggest anyone doing real-time audio from a pc run this little free app...
http://www.thesycon.de/deu/latency_check.shtml
Being a Squeezebox user, I don't quite grasp the "problems" encountered by PC users. I've always thought it has more to do with the relationship (usage) of audio drivers from the corresponding streaming software.
In any case, I stream raw PCM to my SB (not FLAC) as I have a true server, so I hear no "problems".
As for the difference between AAC and WAV : differences become minimal from 400VBR. At 256VBR I can still detect frequent differences. And I wouldn't pay the same price for an inferior content. That's why I don't use sub-256 coding for critical listening.
But, I guess, each has to decide for himself. I decided to stop buying 128 and 256 AAC two years ago. ;)
The results of the poll are in.
Most folk voted the wrong way!
http://www.zerogain.com/forum/showthread.php?t=25890&page=4
http://www.pinkfishmedia.net/forum/showthread.php?t=83892&page=5
Hearing the difference between an AAC file at 256kb/s and its parent WAW file is not easy!
Alex Bell
20-08-2010, 21:19
I always liked WAV better, now I know why
:)
Hi Alex,
Welcome to AOS! :)
Could you please pop into the Welcome area and introduce yourself to our community by supplying your basic geographical location, system details and music tastes, as this is a requirement for all new members joining AOS.
Cheers! :cool:
Marco.
twelvebears
21-08-2010, 09:09
Why AAC ?
Has this got something to do with iTunes shop ?
Why on earth don't iTunes do the decent thing and admit that FLAC is a valid format ?
More to the point, why don't they support it ?
Or at least start offering their own bloody lossless format which all iPods are happy to play already.
Spent some great time listening to Apple Lossless files on my Nano via these (http://www.stereophile.com/headphones/etymotic_research_hf2_amp_hf5_in-ear_headphones/) on holiday and was very pleased with the sound quality.
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.