Post by Michael Messer on May 27, 2009 17:26:29 GMT
I have just been sent a link to an interesting report that is worth a read......there's a lot of it, but it is interesting.
The article is from a hi fi store in Ottawa, Canada.
www.planetofsoundonline.com/articles/compression1.html
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 1.
Being a Mac user and a music lover, when Apple announced the Apple Music Store for digital song downloads I was very excited. The store uses a new codec called AAC to deliver the songs (for 99ยข each). It's a competitor to MP3 and since this is the first major attempt at online music downloads for pay, the format is by default the standard for future services. I wanted to know how good AAC was in comparison to MP3, and finally to see if it could come close to standard CD.
I encoded the same track in iTunes using 96, 128, 160 and 192 kbps AAC and MP3, and one AIFF for reference. The track is from the Kansas City Soundtrack - I Surrender Dear. It's a very well recorded live in the studio jazz piece and it's a track that I know very well. It features a solo saxophone and trumpet which are clearly localized in the mix and very closely resemble the real instruments.
The usual caveats of testing apply. This is an unscientific test and it's not double blind but I think that can be good if you're trying to compare something to a known reference.
96 kbps MP3
Harsh Digititis. Extremely rolled off treble. Very phasey. No dynamics. Horrible overall.
Tonal Accuracy - 5/10
Imaging/Soundstage - 4/10
Naturality - 3/10
Musicality - 3/10
Total - 15/40
96 kbps AAC
Noticeable digital sheen, but overall inoffensive. Very rolled off treble. Poor imaging. Light bass. Slightly phasey.
Tonal Accuracy - 6/10
Imaging/Soundstage - 5/10
Naturality - 5/10
Musicality - 7/10
Total - 23/40
128 kbps MP3
Flat, compressed sound/dynamics. Rolled treble (quite bad). One dimensional, plodding bass.
Tonal Accuracy - 5/10
Imaging/Soundstage - 5/10
Naturality - 4/10
Musicality - 4/10
Total - 18/40
128 kbps AAC
Rolled treble, but not too bad. Light bass especially in transients/impact. Compressed dynamics. Surprisingly musical.
Tonal Accuracy - 7/10
Imaging/Soundstage - 6/10
Naturality - 6/10
Musicality - 8/10
Total - 27/40
160 kbps MP3
Noticeably lighter bass than even lower MP3 bit rates. Smeared/flangey treble.
Tonal Accuracy - 7/10
Imaging/Soundstage - 8/10
Naturality - 7/10
Musicality - 6/10
Total - 28/40
160 kbps AAC
Decent bass weight. Much better treble definition/air. Somewhat compressed dynamics. Good, but still a little lifeless.
Tonal Accuracy - 8/10
Imaging/Soundstage - 8/10
Naturality - 8/10
Musicality - 7/10
Total - 31/40
192 kbps MP3
Midrange somewhat forward. Good imaging.
Tonal Accuracy - 7/10
Imaging/Soundstage - 9/10
Naturality - 9/10
Musicality - 8/10
Total - 33/40
192 kbps AAC
No sparkle. Light bass, although with good detail.
Tonal Accuracy - 7/10
Imaging/Soundstage - 7/10
Naturality - 7/10
Musicality - 7/10
Total - 28/40
AIFF (in comparison)
Beautiful sparkle to piano keys. Generally filled with much more life and atmosphere on a tactile level. Far more musically involving. This is the reference piece so it naturally gets a perfect 40/40 score.
CONCLUSION
On the whole, there weren't any surprises. My observations echo what most people have said about AAC vs. MP3. AAC is higher quality at the same bit rate, so you can use a smaller file to achieve the same quality as MP3 which is a good thing for portable and computer users. Ultimately, both formats still sound pretty bad in their practical ranges compared to CD. I didn't test 256 or 320 kbps because it's impractical for most users to use these encodings. The Apple Music Store for example uses 128 kbps, and if you have room for 320 kbps and you care about sound that much you'll probably use AIFF or just play the CDs themselves.
You'll notice a few anomalies in my findings, such as higher bit rates from the same format getting a lower score, and one case where AAC did worse than MP3. These I attribute first to the interaction between the piece of music chosen in relation to the codecs - sometimes less data sounds better, or more distortion sounds better if it gets the right mix of psychoacoustics. An analogy might be to vinyl or tubes, or even compressed FM radio - getting more of the good part of the music, ie. the fundamentals and less of distracting ambiance/texture can actually sound better, or just more enjoyable sometimes. Secondly, as fidelity increases, flaws, or what's missing can become more apparent. An analogy might be to HDTV. When I first saw HDTV I found it blurry (although much "clearer" than regular TV) because I had jumped exponentially in expectation. I wasn't comparing HDTV to normal TV, my brain had jumped standards to compare it to real life!
*NOTE
I've received tons of e-mail thanking me for the test, and many people have suggested that I use OGG or MP3 with LAME/VBR because they're better than the iTunes standard encoder. I've read many comparisions using different codecs and it's probably true, some are better than MP3 or AAC, but that's not the purpose of my test.
The vast majority of Mac users (and perhaps Windows users soon) use iTunes. Moreover, that's what the Apple Music Store is based on so it makes sense to me to look at what is the defining standard for online music downloads. This is important to keep the bar high for these services in the future. People will always come up with ways to improve their encoded files and these are very legitimate but the majority of people will never even open the preferences.
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 2.
After a very warm response to Part 1 of this test, I decided it was important to do a follow-up test to address many of the suggestions and criticisms. My intention with the original exercise was to compare Apple's implementation of the AAC format to the MP3 standard. Since it is also the format used on the iTunes Music Store which has proved very popular and is now the standard for online music distribution, it was also important to gauge approximately how good the quality of those purchased files was going to be.
The purpose of this test is to compare more codecs at a low, medium and high bit rate. I think from this sampling it will be easy to estimate where other bit rates fit in with various types of music. I chose what is probably the most common hi-fi test track - So What from Miles Davis' Kind of Blue. Besides being very well recorded, it is also easily accessible and I encourage you to repeat the test for yourself to form your own opinions. The tracks were reconverted to AIF using Quicktime, burned to CD using Toast, and tested on my home stereo as in the first test. Here are my results:
(title shows bit rate, format - application - setting)
(ratings are first non-blind, then blind, both out of 10 except total, out of 40)
128 kbps MP3 - iTunes - standard
Bloated bass, no treble, very flat, no dynamics, very grainy cymbals, wishy washy imaging.
Tonal Accuracy - 4, 5
Imaging/Soundstage - 3, 2
Naturality - 3, 2
Musicality - 6, 4
Total - 16 (normal) | 14 (blind test)
128 kbps AAC - iTunes - standard
Unlocalized bloated bass, very rolled treble, cymbals very muted, timid, very flat. Absolutely horrible.
Tonal Accuracy - 4, 5
Imaging/Soundstage - 2, 4
Naturality - 1, 2
Musicality - 1, 3
Total - 8 | 14
128 kbps AAC - QuickTime - best
Much better treble, pleasing imaging but still rolled treble. Not bad overall.
Tonal Accuracy - 6, 7
Imaging/Soundstage - 5, 7
Naturality - 4, 7
Musicality - 6, 8
Total - 21 | 29
128 kbps MP3 - NMP3 - constant bit rate, lame
Bloated bass, too much low-mids, lack of bass definition. Rolled treble. Midrange acceptable.
Tonal Accuracy - 4, 6
Imaging/Soundstage - 6, 5
Naturality - 4, 2
Musicality - 4, 2
Total - 18 | 15
128 kbps OGG - OggDrop - variable bit rate
Better high treble, good bass quality and definition. Some sparkle even. Let down by some very odd phase effects however.
Tonal Accuracy - 7, 8
Imaging/Soundstage - 4, 6
Naturality - 6, 8
Musicality - 6, 8
Total - 23 | 30
192 kbps MP3 - iTunes - standard
Light bass, rather flat and mechanical. Better treble and imaging than previous but still very closed in.
Tonal Accuracy - 6, 7
Imaging/Soundstage - 5, 6
Naturality - 5, 6
Musicality - 5, 5
Total - 21 | 24
192 kbps AAC - iTunes - standard
More natural, but still something opaque about the whole thing.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 6, 8
Naturality - 7, 7
Musicality - 6, 6
Total - 26 | 28
192 kbps AAC - QuickTime - best
Bass a little light, pretty good imaging.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 6
Naturality - 8, 7
Musicality - 7, 5
Total - 29 | 25
196 kbps MP3 - NMP3 - vbr, best
First nice piano timbre. Bass very even and detailed. Good overall detail. Still a little closed but nice extended treble.
Tonal Accuracy - 8, 8
Imaging/Soundstage - 8, 8
Naturality - 8, 8
Musicality - 8, 8
Total - 32 | 32
192 kbps OGG - OggDrop - vbr
Bass too light, not defined. No sparkle but still musical. Some weird treble artifacts like white noise.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 8
Naturality - 6, 7
Musicality - 8, 7
Total - 28 | 29
320 kbps MP3 - iTunes - standard
Bass rather one-notey. Good imaging but flat dynamics. A little uninvolving. Better treble detail than previous.
Tonal Accuracy - 8, 7
Imaging/Soundstage - 7, 7
Naturality - 6, 6
Musicality - 7, 4
Total - 28 | 24
320 kbps AAC - iTunes - standard
Light bass, poor definition. Pretty good treble. Dynamics good. Fairly involving.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 7
Naturality - 7, 6
Musicality - 7, 2
Total - 28 | 22
AIFF (in comparison)
It's quite plain to hear the difference between any of these formats and the original. In the blind test I didn't even listen through the whole song before I was convinced I knew the original and it turned out to be the second track I heard. An obvious cue comes from an unlikely place - the tape hiss. In the original it is rather ungloriously extended into the upper reaches of the treble. On all the compressed formats it sounds simply chopped off. The same goes for all the instruments. The top end that gives instruments life, specifically the piano is just gone so you lose much of the tactility that should be giving you that live feeling.
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 3.
In Part 1 and Part 2 of this test I looked at the relative quality of AAC to MP3 using various settings and bitrates for each of the compressed formats. To make a long story short none of the "standard" compression means sounded very good and only through more time consuming and advanced methods could you get the formats to sound acceptable.
Since I did those two tests a lot has changed in the online music world. The iPod is now the default portable music player and the iTunes Music Store is the default place to get music online. The store uses 128 kbit AAC to encode its tracks so many people now are making significant investments in AAC as a future standard. In my previous tests, I found AAC to be lacking in many ways so I thought it was important to see how it sounded now that people were paying for it. I also was curious if I could hear a difference between Apple's professionally encoded tracks and the same track encoded in iTunes on my Powerbook G4 400. They make claims to be using higher resolution masters in certain cases and many have suggested that they have a more robust encoding algorithm.
To encode the songs I first ripped an AIFF in iTunes (4.7 (41) Mac, Quicktime 6.5.2) then converted to either 128 or 192 AAC.
The three songs I chose were:
1. Brian Wilson - Heroes and Villains - Smile 2004
2. Leon Redbone - I Hate a Man Like You - Champagne Charlie
3. Miles Davis - So What - Kind of Blue
You can go buy these tracks and listen for yourself.
Here are my results:
BRIAN WILSON - HEROES AND VILLAINS - SMILE 2004
This track has a dense layering and a lot of percussive sounds.
iTunes Music Store - 128 kbps AAC
Lack of mid-bass weight and "blossoming" warmth. This track has a lot of symphonics and they were rendered fairly cold compared to the CD. Transients and cymbals were harsher and more metallic.
Tonal Accuracy - 4
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 17.5/20
iTunes Encoder - 128 kbps AAC
For some reason this track sounded worse than the same bit rate from the iTunes Music Store. It simply sounded flat.
Tonal Accuracy - 4
Imaging/Soundstage - 4
Naturality - 4
Musicality - 4
Total - 16/20
iTunes Encoder - 192 kbps AAC
Much better than the 128 tracks and very close to the CD. Only a slight flatness seemed to divide the two.
Tonal Accuracy - 4.5
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 18/20
LEON REDBONE - I HATE A MAN LIKE YOU - CHAMPAGNE CHARLIE
This track has very few instruments and is close-miked. The guitar is one of the best I have heard and the strings can come alive with that tangy metallic sound that you feel in real life. On a good stereo the room can be energized in a similar fashion to the real thing.
iTunes Music Store - 128 kbps AAC
This bit rate stumbled quite noticeably on this track. The sound was flattened so that the depth and nuance of the guitar was lessened.
Tonal Accuracy - 4.5
Imaging/Soundstage - 3.5
Naturality - 4
Musicality - 4
Total - 15.5/20
iTunes Encoder - 128 kbps AAC
This track sounded nearly identical to the one I bought from the Apple Music store. There might have been a little less treble definition.
Tonal Accuracy - 4
Imaging/Soundstage - 3.5
Naturality - 4
Musicality - 4
Total - 15/20
iTunes Encoder - 192 kbps AAC
As with the last song, really quite good. Just a hint of flattening.
Tonal Accuracy - 4.5
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 18/20
MILES DAVIS - SO WHAT - KIND OF BLUE
I decided to carry this track over from the last test because it is so well known and so tough for compressed formats to recreate. The soundstage is huge and the detail in the instruments is almost infinitely layered so it becomes a real chore to dig it all out. As always, the tape hiss presents an interesting standard of its own.
iTunes Music Store - 128 kbps AAC
This encoding really showed the treble failings of a low bit rate. As I found before, it was like the whole top end was chopped off.
Tonal Accuracy - 3.5
Imaging/Soundstage - 4
Naturality - 4
Musicality - 4
Total - 15.5/20
iTunes Encoder - 128 kbps AAC
If the official Apple track was bad, this one seemed to go a step further in losing treble detail and soundstaging.
Tonal Accuracy - 3.5
Imaging/Soundstage - 3.5
Naturality - 3.5
Musicality - 3.5
Total - 14/20
iTunes Encoder - 192 kbps AAC
A much better result than the lower bit rates but still quite noticeably inferior to the original.
Tonal Accuracy - 4
Imaging/Soundstage - 4.5
Naturality - 4
Musicality - 4.5
Total - 17/20
CONCLUSION
Overall it seems likely to me that Apple has improved its AAC implementation since I first tested it. If someone from the company wants to outline any general changes I would be curious to know but with these songs, to my ears, it sounds better.
It's also interesting that the files from the Apple iTunes Music Store seem to have a very slight edge over the ones I encoded on my home computer.
As I pointed out before, I think a test like this will be affected to a large degree by the songs that are chosen. I tried to pick a range of things that might give the encoders difficulty. On the whole I would conclude that well recorded tracks with exceptional detail come out the worst when encoded. This follows from common sense but it should give you an idea of how good the music you like might sound.
Is it good enough to pay for? You must answer that question yourself. For me, losing 10-25% of the musical information is like losing that much enjoyment and if buying CDs or LPs is as affordable I'm willing to buy those formats to not lose out on anything. For those of you who want to walk the middle ground, I was pleasantly surprised to see that 192 kbps AAC is sounding very good. If you have an iPod I would have no reservations about encoding all my music at that rate. Hopefully as time goes on Apple will do the same.
Shine On
Michael.
The article is from a hi fi store in Ottawa, Canada.
www.planetofsoundonline.com/articles/compression1.html
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 1.
Being a Mac user and a music lover, when Apple announced the Apple Music Store for digital song downloads I was very excited. The store uses a new codec called AAC to deliver the songs (for 99ยข each). It's a competitor to MP3 and since this is the first major attempt at online music downloads for pay, the format is by default the standard for future services. I wanted to know how good AAC was in comparison to MP3, and finally to see if it could come close to standard CD.
I encoded the same track in iTunes using 96, 128, 160 and 192 kbps AAC and MP3, and one AIFF for reference. The track is from the Kansas City Soundtrack - I Surrender Dear. It's a very well recorded live in the studio jazz piece and it's a track that I know very well. It features a solo saxophone and trumpet which are clearly localized in the mix and very closely resemble the real instruments.
The usual caveats of testing apply. This is an unscientific test and it's not double blind but I think that can be good if you're trying to compare something to a known reference.
96 kbps MP3
Harsh Digititis. Extremely rolled off treble. Very phasey. No dynamics. Horrible overall.
Tonal Accuracy - 5/10
Imaging/Soundstage - 4/10
Naturality - 3/10
Musicality - 3/10
Total - 15/40
96 kbps AAC
Noticeable digital sheen, but overall inoffensive. Very rolled off treble. Poor imaging. Light bass. Slightly phasey.
Tonal Accuracy - 6/10
Imaging/Soundstage - 5/10
Naturality - 5/10
Musicality - 7/10
Total - 23/40
128 kbps MP3
Flat, compressed sound/dynamics. Rolled treble (quite bad). One dimensional, plodding bass.
Tonal Accuracy - 5/10
Imaging/Soundstage - 5/10
Naturality - 4/10
Musicality - 4/10
Total - 18/40
128 kbps AAC
Rolled treble, but not too bad. Light bass especially in transients/impact. Compressed dynamics. Surprisingly musical.
Tonal Accuracy - 7/10
Imaging/Soundstage - 6/10
Naturality - 6/10
Musicality - 8/10
Total - 27/40
160 kbps MP3
Noticeably lighter bass than even lower MP3 bit rates. Smeared/flangey treble.
Tonal Accuracy - 7/10
Imaging/Soundstage - 8/10
Naturality - 7/10
Musicality - 6/10
Total - 28/40
160 kbps AAC
Decent bass weight. Much better treble definition/air. Somewhat compressed dynamics. Good, but still a little lifeless.
Tonal Accuracy - 8/10
Imaging/Soundstage - 8/10
Naturality - 8/10
Musicality - 7/10
Total - 31/40
192 kbps MP3
Midrange somewhat forward. Good imaging.
Tonal Accuracy - 7/10
Imaging/Soundstage - 9/10
Naturality - 9/10
Musicality - 8/10
Total - 33/40
192 kbps AAC
No sparkle. Light bass, although with good detail.
Tonal Accuracy - 7/10
Imaging/Soundstage - 7/10
Naturality - 7/10
Musicality - 7/10
Total - 28/40
AIFF (in comparison)
Beautiful sparkle to piano keys. Generally filled with much more life and atmosphere on a tactile level. Far more musically involving. This is the reference piece so it naturally gets a perfect 40/40 score.
CONCLUSION
On the whole, there weren't any surprises. My observations echo what most people have said about AAC vs. MP3. AAC is higher quality at the same bit rate, so you can use a smaller file to achieve the same quality as MP3 which is a good thing for portable and computer users. Ultimately, both formats still sound pretty bad in their practical ranges compared to CD. I didn't test 256 or 320 kbps because it's impractical for most users to use these encodings. The Apple Music Store for example uses 128 kbps, and if you have room for 320 kbps and you care about sound that much you'll probably use AIFF or just play the CDs themselves.
You'll notice a few anomalies in my findings, such as higher bit rates from the same format getting a lower score, and one case where AAC did worse than MP3. These I attribute first to the interaction between the piece of music chosen in relation to the codecs - sometimes less data sounds better, or more distortion sounds better if it gets the right mix of psychoacoustics. An analogy might be to vinyl or tubes, or even compressed FM radio - getting more of the good part of the music, ie. the fundamentals and less of distracting ambiance/texture can actually sound better, or just more enjoyable sometimes. Secondly, as fidelity increases, flaws, or what's missing can become more apparent. An analogy might be to HDTV. When I first saw HDTV I found it blurry (although much "clearer" than regular TV) because I had jumped exponentially in expectation. I wasn't comparing HDTV to normal TV, my brain had jumped standards to compare it to real life!
*NOTE
I've received tons of e-mail thanking me for the test, and many people have suggested that I use OGG or MP3 with LAME/VBR because they're better than the iTunes standard encoder. I've read many comparisions using different codecs and it's probably true, some are better than MP3 or AAC, but that's not the purpose of my test.
The vast majority of Mac users (and perhaps Windows users soon) use iTunes. Moreover, that's what the Apple Music Store is based on so it makes sense to me to look at what is the defining standard for online music downloads. This is important to keep the bar high for these services in the future. People will always come up with ways to improve their encoded files and these are very legitimate but the majority of people will never even open the preferences.
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 2.
After a very warm response to Part 1 of this test, I decided it was important to do a follow-up test to address many of the suggestions and criticisms. My intention with the original exercise was to compare Apple's implementation of the AAC format to the MP3 standard. Since it is also the format used on the iTunes Music Store which has proved very popular and is now the standard for online music distribution, it was also important to gauge approximately how good the quality of those purchased files was going to be.
The purpose of this test is to compare more codecs at a low, medium and high bit rate. I think from this sampling it will be easy to estimate where other bit rates fit in with various types of music. I chose what is probably the most common hi-fi test track - So What from Miles Davis' Kind of Blue. Besides being very well recorded, it is also easily accessible and I encourage you to repeat the test for yourself to form your own opinions. The tracks were reconverted to AIF using Quicktime, burned to CD using Toast, and tested on my home stereo as in the first test. Here are my results:
(title shows bit rate, format - application - setting)
(ratings are first non-blind, then blind, both out of 10 except total, out of 40)
128 kbps MP3 - iTunes - standard
Bloated bass, no treble, very flat, no dynamics, very grainy cymbals, wishy washy imaging.
Tonal Accuracy - 4, 5
Imaging/Soundstage - 3, 2
Naturality - 3, 2
Musicality - 6, 4
Total - 16 (normal) | 14 (blind test)
128 kbps AAC - iTunes - standard
Unlocalized bloated bass, very rolled treble, cymbals very muted, timid, very flat. Absolutely horrible.
Tonal Accuracy - 4, 5
Imaging/Soundstage - 2, 4
Naturality - 1, 2
Musicality - 1, 3
Total - 8 | 14
128 kbps AAC - QuickTime - best
Much better treble, pleasing imaging but still rolled treble. Not bad overall.
Tonal Accuracy - 6, 7
Imaging/Soundstage - 5, 7
Naturality - 4, 7
Musicality - 6, 8
Total - 21 | 29
128 kbps MP3 - NMP3 - constant bit rate, lame
Bloated bass, too much low-mids, lack of bass definition. Rolled treble. Midrange acceptable.
Tonal Accuracy - 4, 6
Imaging/Soundstage - 6, 5
Naturality - 4, 2
Musicality - 4, 2
Total - 18 | 15
128 kbps OGG - OggDrop - variable bit rate
Better high treble, good bass quality and definition. Some sparkle even. Let down by some very odd phase effects however.
Tonal Accuracy - 7, 8
Imaging/Soundstage - 4, 6
Naturality - 6, 8
Musicality - 6, 8
Total - 23 | 30
192 kbps MP3 - iTunes - standard
Light bass, rather flat and mechanical. Better treble and imaging than previous but still very closed in.
Tonal Accuracy - 6, 7
Imaging/Soundstage - 5, 6
Naturality - 5, 6
Musicality - 5, 5
Total - 21 | 24
192 kbps AAC - iTunes - standard
More natural, but still something opaque about the whole thing.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 6, 8
Naturality - 7, 7
Musicality - 6, 6
Total - 26 | 28
192 kbps AAC - QuickTime - best
Bass a little light, pretty good imaging.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 6
Naturality - 8, 7
Musicality - 7, 5
Total - 29 | 25
196 kbps MP3 - NMP3 - vbr, best
First nice piano timbre. Bass very even and detailed. Good overall detail. Still a little closed but nice extended treble.
Tonal Accuracy - 8, 8
Imaging/Soundstage - 8, 8
Naturality - 8, 8
Musicality - 8, 8
Total - 32 | 32
192 kbps OGG - OggDrop - vbr
Bass too light, not defined. No sparkle but still musical. Some weird treble artifacts like white noise.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 8
Naturality - 6, 7
Musicality - 8, 7
Total - 28 | 29
320 kbps MP3 - iTunes - standard
Bass rather one-notey. Good imaging but flat dynamics. A little uninvolving. Better treble detail than previous.
Tonal Accuracy - 8, 7
Imaging/Soundstage - 7, 7
Naturality - 6, 6
Musicality - 7, 4
Total - 28 | 24
320 kbps AAC - iTunes - standard
Light bass, poor definition. Pretty good treble. Dynamics good. Fairly involving.
Tonal Accuracy - 7, 7
Imaging/Soundstage - 7, 7
Naturality - 7, 6
Musicality - 7, 2
Total - 28 | 22
AIFF (in comparison)
It's quite plain to hear the difference between any of these formats and the original. In the blind test I didn't even listen through the whole song before I was convinced I knew the original and it turned out to be the second track I heard. An obvious cue comes from an unlikely place - the tape hiss. In the original it is rather ungloriously extended into the upper reaches of the treble. On all the compressed formats it sounds simply chopped off. The same goes for all the instruments. The top end that gives instruments life, specifically the piano is just gone so you lose much of the tactility that should be giving you that live feeling.
REPORT ON THE RELATIVE QUALITY OF AAC AUDIO TO MP3 PART 3.
In Part 1 and Part 2 of this test I looked at the relative quality of AAC to MP3 using various settings and bitrates for each of the compressed formats. To make a long story short none of the "standard" compression means sounded very good and only through more time consuming and advanced methods could you get the formats to sound acceptable.
Since I did those two tests a lot has changed in the online music world. The iPod is now the default portable music player and the iTunes Music Store is the default place to get music online. The store uses 128 kbit AAC to encode its tracks so many people now are making significant investments in AAC as a future standard. In my previous tests, I found AAC to be lacking in many ways so I thought it was important to see how it sounded now that people were paying for it. I also was curious if I could hear a difference between Apple's professionally encoded tracks and the same track encoded in iTunes on my Powerbook G4 400. They make claims to be using higher resolution masters in certain cases and many have suggested that they have a more robust encoding algorithm.
To encode the songs I first ripped an AIFF in iTunes (4.7 (41) Mac, Quicktime 6.5.2) then converted to either 128 or 192 AAC.
The three songs I chose were:
1. Brian Wilson - Heroes and Villains - Smile 2004
2. Leon Redbone - I Hate a Man Like You - Champagne Charlie
3. Miles Davis - So What - Kind of Blue
You can go buy these tracks and listen for yourself.
Here are my results:
BRIAN WILSON - HEROES AND VILLAINS - SMILE 2004
This track has a dense layering and a lot of percussive sounds.
iTunes Music Store - 128 kbps AAC
Lack of mid-bass weight and "blossoming" warmth. This track has a lot of symphonics and they were rendered fairly cold compared to the CD. Transients and cymbals were harsher and more metallic.
Tonal Accuracy - 4
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 17.5/20
iTunes Encoder - 128 kbps AAC
For some reason this track sounded worse than the same bit rate from the iTunes Music Store. It simply sounded flat.
Tonal Accuracy - 4
Imaging/Soundstage - 4
Naturality - 4
Musicality - 4
Total - 16/20
iTunes Encoder - 192 kbps AAC
Much better than the 128 tracks and very close to the CD. Only a slight flatness seemed to divide the two.
Tonal Accuracy - 4.5
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 18/20
LEON REDBONE - I HATE A MAN LIKE YOU - CHAMPAGNE CHARLIE
This track has very few instruments and is close-miked. The guitar is one of the best I have heard and the strings can come alive with that tangy metallic sound that you feel in real life. On a good stereo the room can be energized in a similar fashion to the real thing.
iTunes Music Store - 128 kbps AAC
This bit rate stumbled quite noticeably on this track. The sound was flattened so that the depth and nuance of the guitar was lessened.
Tonal Accuracy - 4.5
Imaging/Soundstage - 3.5
Naturality - 4
Musicality - 4
Total - 15.5/20
iTunes Encoder - 128 kbps AAC
This track sounded nearly identical to the one I bought from the Apple Music store. There might have been a little less treble definition.
Tonal Accuracy - 4
Imaging/Soundstage - 3.5
Naturality - 4
Musicality - 4
Total - 15/20
iTunes Encoder - 192 kbps AAC
As with the last song, really quite good. Just a hint of flattening.
Tonal Accuracy - 4.5
Imaging/Soundstage - 4.5
Naturality - 4.5
Musicality - 4.5
Total - 18/20
MILES DAVIS - SO WHAT - KIND OF BLUE
I decided to carry this track over from the last test because it is so well known and so tough for compressed formats to recreate. The soundstage is huge and the detail in the instruments is almost infinitely layered so it becomes a real chore to dig it all out. As always, the tape hiss presents an interesting standard of its own.
iTunes Music Store - 128 kbps AAC
This encoding really showed the treble failings of a low bit rate. As I found before, it was like the whole top end was chopped off.
Tonal Accuracy - 3.5
Imaging/Soundstage - 4
Naturality - 4
Musicality - 4
Total - 15.5/20
iTunes Encoder - 128 kbps AAC
If the official Apple track was bad, this one seemed to go a step further in losing treble detail and soundstaging.
Tonal Accuracy - 3.5
Imaging/Soundstage - 3.5
Naturality - 3.5
Musicality - 3.5
Total - 14/20
iTunes Encoder - 192 kbps AAC
A much better result than the lower bit rates but still quite noticeably inferior to the original.
Tonal Accuracy - 4
Imaging/Soundstage - 4.5
Naturality - 4
Musicality - 4.5
Total - 17/20
CONCLUSION
Overall it seems likely to me that Apple has improved its AAC implementation since I first tested it. If someone from the company wants to outline any general changes I would be curious to know but with these songs, to my ears, it sounds better.
It's also interesting that the files from the Apple iTunes Music Store seem to have a very slight edge over the ones I encoded on my home computer.
As I pointed out before, I think a test like this will be affected to a large degree by the songs that are chosen. I tried to pick a range of things that might give the encoders difficulty. On the whole I would conclude that well recorded tracks with exceptional detail come out the worst when encoded. This follows from common sense but it should give you an idea of how good the music you like might sound.
Is it good enough to pay for? You must answer that question yourself. For me, losing 10-25% of the musical information is like losing that much enjoyment and if buying CDs or LPs is as affordable I'm willing to buy those formats to not lose out on anything. For those of you who want to walk the middle ground, I was pleasantly surprised to see that 192 kbps AAC is sounding very good. If you have an iPod I would have no reservations about encoding all my music at that rate. Hopefully as time goes on Apple will do the same.
Shine On
Michael.