Just about everywhere I've looked on the net, people have been talking about how the new iTunes 7.5/QT 7.3 AAC encoder is an noticeable improvement over the iTunes 7.3/QT 7.2 AAC encoder. This has not been my experience at all at the 128 VBR bitrate. A good example (and I have more if one is not sufficient) from a CD that I expect many people have is "What Do You Do for Money Honey" from AC/DC's "Back in Black".

Here is my ABX test between my WavPack image and a 128 VBR encode using iTunes 7.3.1.3 / QT 7.2 for windows:



Here is my ABX test between my WavPack image and a 128 VBR encode using iTunes 7.5.0.2.0 / QT 7.3 for windows:



The problems I'm hearing exemplified in the first 10 seconds of the above track are the same cymbal smearing and distortion problems that drove me from iTunes MP3 of old to AAC in the first place (all those years ago).

Is anyone else hearing this? Am I the only one having serious quality problems with the new AAC encoder in iTunes 7.5/QT 7.3?

EDIT: Some quick testing has indicated that iTunes 7.6 / QuickTime 7.4 have done nothing to correct the encoder problems I'm hearing.

Now that iTunes and QuickTime have been updated once again, I figured I'd do another test. This time the song is "Watershed" from the Foo Fighters album "Foo Fighters".

Here is my ABX test between my WavPack image and a 128 VBR encode using iTunes 7.3.1.3 / QT 7.2 for windows:



Here is my ABX test between my WavPack image and a 128 VBR encode using iTunes 7.6.0.2.9 / QT 7.4 for windows:



So... I'm still waiting for a fix to make iTunes AAC function as well at 128 VBR as it did before the last encoder update.

No apparent changes with iTunes 7.6.2.9 / QuickTime 7.4.5, i.e. it still sounds like garbage compared to the previous version.

hey dbAmp,

As a disclaimer, this is my first post on hyrdogenaudio, and my first abx test as well. I'm surprised no one else has replied to this thread yet, because this lapse in quality of the Quicktime AAC encoder seems like a big deal (at least to me anyway).

My process: converted my FLAC file to WAV using foobar. encoded the WAV in iTunes to 128 kbps (no vbr). used foobar to abx the two samples. I used bose around-ear headphones. I get the same results as you:



Anyhow, I was rather surprised that I was able to abx this so easy seeing that I have no experience with this sort of thing. But the problem was pretty easy to identify, especially when the 2nd guitar (I think) comes in at the 10 second mark. Does apple know about this issue?

Apple doesn't know about issues unless they're reported.

I'm sure they don't unless they browse these forums like some of the Nero devs, but I guess what I meant was...has dbAmp or anyone else reported this issue?

Also, I was using iTunes 7.6.2.9 and Quicktime 7.4.5. I've checked some other bitrates too. at 96 kbps the problem is much more apparent, and I was able to abx the problem at 192 as well. I have not tried going higher than that.

I have notified Apple with my ABX results using a song by Bleeding Through called For Love And Failing off of their "The Truth" album. I also tested a song by A Perfect Murder called War Of Aggression off of their "War Of Aggression" album. I pretty much get the same results so I don't feel the need to post the code. I can post the ABX results though as I still have them in my clipboard. I used my Bose TriPort (the re-design) headphones that are about 2 years old. They aren't the best but the issue is there even with those headphones.

And Nero developers would like to hear how does Nero AAC Encoder handle these samples

I actually was contacted by someone on the forums from Apple and exchanged a few e-mails. His response to my above test was:

"Thanks for sending it to me. We can hear some minor distortion, however, I am sorry to say it's very minor to us, so it's hard for us to justify some time to fix it. The new encoder has been well tuned internally. Based on our internal objective and subjective listening tests, the new encoder outperforms the old one by about 10%, that is, if keeps the same sound quality, then the bitrate can be reduced by 10%, or if at the same bitrate, the sound quality of the new encoder is obviously better than the old one over a broad range of our music collection. There might be some very minor regression in some particular area or certain song as you may have encountered one, which is expected as this is a lossy codec..."

I think this just means my ears must be more subjective than most...

I found these "minor regressions" frequent and serious enough that I stopped using the iTunes AAC codec completely (I've been able to ABX problem tracks even at high bit rates). As to jenzo's surprise... I don't think of myself as someone with golden ears and I'm not using a pro setup (Dell Inspiron E1505 and Sennheiser PX100 headphones). That being said, I don't know the conditions under which Apple is testing. It may be that they are trying to optimize for people using an iPod with stock earbuds (that would be my goal if I were Apple).

If these problems are widespread, I trust that Apple will address them eventually. I would therefore encourage anyone following this thread to find more problem samples and post their ABX results here.



Which version of Nero and at what settings would you like me to test?

For now please take a look at this thread:
http://www.hydrogenaudio.org/forums/index....=14&t=61724
We might release test version soon, but we still need to investigate problems in August version. It takes time because it is very hard to get rid of some problems without introducing new ones. Additionally we didn't see to much interest in that thread for helping us improve and release new version .

I looked at the above thread, but since I've never really used the Nero encoder before, I'm not sure what command line would be appropriate. Further, I've never used the February version, so I'm not sure what problems people were hearing after switching to the August version.

I did test the above problem tracks using the August version and default foobar2000 command line and -lc with quality levels of .35 and .40. I could ABX each at .35 and none of them at .40. I hope this helps, even though I get the feeling that's not quite the testing you wanted.

You should be happy to know, that even at .35, IMHO, the problems were much less obvious than the glaring artifacts now present in the iTunes AAC encoder at 128 VBR.