Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Is iTunes' AAC Distinguishable From Uncompressed Audio?

stridey

stridey

macrumors 65816
Original poster
Jan 21, 2005
1,136
0
Massachusetts, Connecticut
Hi guys,

I've written up an article over on my blog about why I don't think AAC is distinguishable from uncompressed AIFF audio. Now, before you audiophiles get too bristled up, read the article. ;) It includes sound tests, in which I'm totally unable to hear the difference.

Check it out

Also, if you like it, I'd love to Get some diggs on it. ;)
 
shadowmoses

shadowmoses

macrumors 68000
Mar 6, 2005
1,821
0
very cool article, you been dugg, hopefully you'll get some front page action on digg.com

ShadoW
 
Eidorian

Eidorian

macrumors Penryn
Mar 23, 2005
29,190
386
Indianapolis
Very nice article. I could barely tell the transition using my iMac's internal speakers but it was there. It was just some minor frequency issues in the very high end. I guess you'd need some better ear gear than that to REALLY notice it.
 
Nermal

Nermal

Moderator
Staff member
Dec 7, 2002
20,651
4,052
New Zealand
I've found that it depends on the music. Some songs sound perfect in AAC, and some sound terrible.
 
A

AP_piano295

macrumors 65816
Mar 9, 2005
1,076
17
Eidorian said:
Very nice article. I could barely tell the transition using my iMac's internal speakers but it was there. It was just some minor frequency issues in the very high end. I guess you'd need some better ear gear than that to REALLY notice it.
Click to expand...

Really because I listened to them all on sure i4c's about five times and i notice absolutely no variation. I doubt you could ever pick up such minor if exsistant variations on your iMac's internals.
 
WinterMute

WinterMute

Moderator emeritus
Jan 19, 2003
4,776
5
London, England
I've commented on this before at MR, the last time you brought it to our attention...;)


Precis:

You need trained ears to hear and a more sensetive system to listen on.

AAC (and pretty much all compressed lossy formats) are clearly different to uncompressed PCM (aiff or wav) files, which are, in turn, different to source analog or HD audio masters.

I personally would not be able to tell the difference between a picture shot with one brand of camera to that of another, unless I was taught how to do so, this does not mean the differences don't exist, it means I can't detect them. Others can.

Anyone who is a professional in their field is going to be more sensitive to the variations of technique employed than even the most informed amateur, this may be one definition of "professional".

AAC is probably the most musical of the lossy codecs (it's on my iPod), but it cannot re-create the original waveform, hence the term "lossy".

Lossless encoding is a different and far more problematic issue, because to my ears they sound different, but report an identical waveform. Go figure.
 
zimv20

zimv20

macrumors 601
Jul 18, 2002
4,402
11
toronto
the blog said:
The AAC I used is 160 Kbps
Click to expand...
that sample rate is near where i have a hard time telling the difference. i've sometimes heard differences at 192 kbps, but have not heard a difference at 256 kbps.

all my listening tests were done in my partially-treated room. perhaps on better monitors and in a better room i'd do better.
 
V

vniow

macrumors G4
Jul 18, 2002
10,266
1
I accidentally my whole location.
stridey said:
Also, if you like it, I'd love to Get some diggs on it. ;)
Click to expand...

Lurker?

hmm.PNG
 
emw

emw

macrumors G4
Aug 2, 2004
11,172
0
Interesting article. I'd have to agree that in the majority of situations, even if there is a difference, it won't matter. Too much external noise interference on planes, trains, etc. Even when I listen at home on a nicer setup, I'm usually not listening intently to the music; rather I enjoy it as background entertainment.
 
stridey

stridey

macrumors 65816
Original poster
Jan 21, 2005
1,136
0
Massachusetts, Connecticut
WinterMute said:
~snip
You need trained ears to hear and a more sensetive system to listen on.

AAC (and pretty much all compressed lossy formats) are clearly different to uncompressed PCM (aiff or wav) files, which are, in turn, different to source analog or HD audio masters.

I personally would not be able to tell the difference between a picture shot with one brand of camera to that of another, unless I was taught how to do so, this does not mean the differences don't exist, it means I can't detect them. Others can.

Anyone who is a professional in their field is going to be more sensitive to the variations of technique employed than even the most informed amateur, this may be one definition of "professional".

AAC is probably the most musical of the lossy codecs (it's on my iPod), but it cannot re-create the original waveform, hence the term "lossy".
Click to expand...

I agree that AAC is inherently lossy. My argument is that you can't actually hear the difference. I'm interested in knowing; if you consider yourself to be a professional in audio (I don't know), can you hear the difference in my sample snippets?

I've had a lot of people claim to hear the difference, but nobody yet has even been able to identify with accuracy which pieces begin lossy and end uncompressed, and which do the reverse.
 
beatsme

beatsme

macrumors 65816
Oct 6, 2005
1,204
2
actually...

I didn't listen to the side-by-sides you presented, but I can sometimes tell the difference between .aif and AAC, though only through actual speakers and definitely not through any of the headphones I have. I think it kinda depends on what music you're listening to. Coltrane, for example, sounds a little fuller (to my ears) in .aif than AAC...ditto Miles Davis. And some classical can be (to me) a little thin. For straight rock, I can't tell any difference.

and yes, I am the guy that still says vinyl sounds better than anything...
 
stridey

stridey

macrumors 65816
Original poster
Jan 21, 2005
1,136
0
Massachusetts, Connecticut
beatsme said:
I didn't listen to the side-by-sides you presented, but I can sometimes tell the difference between .aif and AAC, though only through actual speakers and definitely not through any of the headphones I have. I think it kinda depends on what music you're listening to. Coltrane, for example, sounds a little fuller (to my ears) in .aif than AAC...ditto Miles Davis. And some classical can be (to me) a little thin. For straight rock, I can't tell any difference.

and yes, I am the guy that still says vinyl sounds better than anything...
Click to expand...

See, here's the thing. I know you think you've heard the difference. I've thought I heard the difference too, but I always knew whether it was AIFF or AAC before I decided that it was better/worse. Somebody (in the digg article?) mentioned that listening to AAC starts "tiring out" your ears after a long time, and I think I might buy that. I certainly buy that you *think* you're hearing a difference, but I don't think I buy that you really are, until you conduct a blind "taste" test. Various scientific studies over the decades point very strongly to the "placebo effect" being quite powerful; powerful enough to make us really believe what our mind wants us to believe.

And yeah, I think vinyl sounds better than anything too, but not because it's higher quality. Just because it 'feels' good. (childhood connotations maybe? nostalgia? I like the warmth? who knows).

I do suggest checking out the article/sound samples. I go into a lot more depth there. ;)

Last addendum: I know I'm making some pretty tall claims, but I've given it a lot of thought, and I'm pretty confident that they're correct. It's still just my opinion, though. :)
 
zimv20

zimv20

macrumors 601
Jul 18, 2002
4,402
11
toronto
stridey said:
I always knew whether it was AIFF or AAC before I decided that it was better/worse.
Click to expand...
after i mix a song and drop the aiff into itunes, i then convert it to either mp3 or aac so i can share it with the band.

i then listen with an ear to "has this conversion lost anything"? Because i just got done mixing it, i'm already very familiar with the full-resolution version, and i want to send out a compressed version which, to my ears, sounds as good.

in this scenario, is it important for me to do a double-blind? i don't think so. if i hear a difference, i re-encode at a higher bitrate. repeat until i can't tell the difference, even knowing which is which.

related, one place to check for compression degradation is in the cymbal tails.
 
WinterMute

WinterMute

Moderator emeritus
Jan 19, 2003
4,776
5
London, England
stridey said:
I agree that AAC is inherently lossy. My argument is that you can't actually hear the difference. I'm interested in knowing; if you consider yourself to be a professional in audio (I don't know), can you hear the difference in my sample snippets?

I've had a lot of people claim to hear the difference, but nobody yet has even been able to identify with accuracy which pieces begin lossy and end uncompressed, and which do the reverse.
Click to expand...

Can you tell me how you made the files please, and what format you are streaming them in, what application did you use that can play back 2 different audio codecs and crossfade between them, not even ProTools will do that, all the Pro audio systems will convert the incoming audio to a set file-type.

I'll need to know the file creator and codec settings for each version and the final encoder that created the files you are using, and I'll need to be able to download them into a system I can play them accurately from, as the speakers on my PowerBook are genuinely useless for audio comparisons.

A better test is to play 2 separate versions of the audio from 2 files through the same delivery system, you can still run blind test (we do it all the time in the Uni and in the MPG, and the AES does "golden ear" listening tests on a lot of new systems regularly).

This gives the audience a much better chance of hearing the differences, which are night and day in AAC/aiff terms.

There are too many unknown variables in your tests, I have no idea what I'm hearing, or even if the effect you describe has survived the transcoding process. The other issue is that the audio clips are far too short and the source files on three of the clips show significant distortion in the original audio recording process, classical music is thought by some to represent a much greater challenge than contemporary music, but you need a couple of minutes of each type to be able to appraise the tonal qualities accurately.

For the record, I'm a 20 year commercial recording engineer and I lecture in audio technology at a London University, I'm also a musician.
 
stridey

stridey

macrumors 65816
Original poster
Jan 21, 2005
1,136
0
Massachusetts, Connecticut
WinterMute said:
Can you tell me how you made the files please, and what format you are streaming them in, what application did you use that can play back 2 different audio codecs and crossfade between them, not even ProTools will do that, all the Pro audio systems will convert the incoming audio to a set file-type.

I'll need to know the file creator and codec settings for each version and the final encoder that created the files you are using, and I'll need to be able to download them into a system I can play them accurately from, as the speakers on my PowerBook are genuinely useless for audio comparisons.

A better test is to play 2 separate versions of the audio from 2 files through the same delivery system, you can still run blind test (we do it all the time in the Uni and in the MPG, and the AES does "golden ear" listening tests on a lot of new systems regularly).

This gives the audience a much better chance of hearing the differences, which are night and day in AAC/aiff terms.

There are too many unknown variables in your tests, I have no idea what I'm hearing, or even if the effect you describe has survived the transcoding process. The other issue is that the audio clips are far too short and the source files on three of the clips show significant distortion in the original audio recording process, classical music is thought by some to represent a much greater challenge than contemporary music, but you need a couple of minutes of each type to be able to appraise the tonal qualities accurately.

For the record, I'm a 20 year commercial recording engineer and I lecture in audio technology at a London University, I'm also a musician.
Click to expand...

Certainly. Here's what I did. I copied the AIFF source file straight from CD. I also imported the source into iTunes (using the encoder options I mentioned in the entry). Once I had the two files, I opened them up in Sound Studio (I needed to use the "import using Quicktime" instead of just "open" for the AAC). I found the place I wanted to splice them on each song (carefully, down to the correct sample), cut, pasted, and was on my merry way. I then saved the completed files using AIFF 44.1 kHz and 16 bits (since it's CD quality). If you do a spectral analysis on the samples, I'm sure you'll be able to see the cut - on one side the frequency will go up to 22 kHz, and on the other side it should vary.

I also attempted to make the cut in aurally obvious places (like the middle of a sustained trumpet note, in the Dizzy piec) - I *wanted* to be able to hear the difference.

Sorry that you don't like the length/samples I picked. I really was trying to make it the most scientific test possible, and figured that in a short sample it woudl be easier to hear the crossover. Still, if you would listen to it, I'll gladly do a couple minutes of a classical piece the same way I did these four samples. :)
 
beatsme

beatsme

macrumors 65816
Oct 6, 2005
1,204
2
somebody tell me if I'm full of it...

I have long thought that the reason vinyl sounds better is because of the mechanism. The needle/record combo works sort of like a music box and sort of like a microphone, right? The groove spins along below the needle, and the needle picks up the vibration from the groove and sends along an electrical pulse (or a series of pulses, actually) to the pre-amp/amp which is then converted into sound.

If I'm wrong about that, somebody say so.

anyway, I think of the needle as being somewhat like a guitar string; it's going to react in different ways depending on how you pluck it. The dynamics of the needle and groove interaction give vinyl a much more "real" sound, since the sustain/rate of decay is an actual thing and not something computer-generated.

I obviously have a lot of free time...
 
zimv20

zimv20

macrumors 601
Jul 18, 2002
4,402
11
toronto
maybe i'm talking out of my ass here, but i'd always thought that vinyl offers a sonic advantage because it hasn't been reduced to 1's and 0's. rather, an actual continual curve is represented in the vinyl, whereas a CD's curve is reconstructed from the information on the CD.

also, any converters performance relies heavily on the analog and other circuitry surrounding the chip. so it may be the case that the turntables you've heard on relying on electronic components such as transformers, and not so much on multi-purpose surface-mount chips.

all imo.
 
H

howesey

macrumors 6502a
Dec 3, 2005
535
0
The equipment you tried them on is not a good way to test it. They are designed to sound crap.

As for AAC and RAW PCM, I can tell a huge difference. I do not care on my ipod while on the train though. I'm the sort of person that can tell the difference between a copied and the orginal CD.

My hi-fi so downstairs and is not connected to any computer so I cannot do a full test. My iMac in my bedroom runs through some crappy sony processor to some B&W 600 series floorstanders.
 
stridey

stridey

macrumors 65816
Original poster
Jan 21, 2005
1,136
0
Massachusetts, Connecticut
howesey said:
The equipment you tried them on is not a good way to test it. They are designed to sound crap.

As for AAC and RAW PCM, I can tell a huge difference. I do not care on my ipod while on the train though. I'm the sort of person that can tell the difference between a copied and the orginal CD.

My hi-fi so downstairs and is not connected to any computer so I cannot do a full test. My iMac in my bedroom runs through some crappy sony processor to some B&W 600 series floorstanders.
Click to expand...

The equipent I tried on them is designed to sound crap? Look, I have no problem with you saying I'm full of s**t about AAC sounding just like AIFF, but those speakers are some of the best damned studio monitors money can buy. Whatever. To each their own. :p
 
zimv20

zimv20

macrumors 601
Jul 18, 2002
4,402
11
toronto
stridey said:
Well, if you say they're no good I'll take your word on it... but others claim that they're very good.
Click to expand...
yeah, of course mackie is going to hype their monitors.

i've got a pair of the 624s, and i think they're actually better than the 824s, which a friend let me borrow for a year. so i know 'em both pretty well.

at best, these monitors are at the bottom of the middle tier, imnsho. i'm saving up for a new pair and looking at monitors from adam, dynaudio, barefoot audio, magnepans and others. i expect any of these to kick the crap out of my 624s.

btw -- my 624s "improved" dramatically when i added some bass traps to my room. before that, everything sounded good (as in, too good) through them. now, at least, songs which are flat sound flat and i've got a better idea of what's going on in the bottom. still not enough for my tastes, though.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom