Nero AAC vs. iTunes AAC (spectrum-wise) Options

[subinbar]

I'm torn between using NeroAACEnc or iTunes to encode my lossless collection for mobile use. Nero uses a lowpass filter, while iTunes Plus does not.

Theoretically, can Nero's lowpass settings affect sound quality at higher bitrates (256kbps)? Or by eliminating higher frequencies that the ears can't hear, does Nero make more bandwidth available to lower frequencies that the ear CAN hear? I do not have the equipment and/or ears necessary to consistently tell the difference between these two encoders when encoding from a lossless source. However, I have found that transcoding from a lossy source, I can usually tell a slight difference in the high frequencies (Nero doesn't preserve as well as iTunes IMO... or Vorbis for that matter)

I took some screenshots of spectrums of these 4 different files-

Original Lossless file (36.2 MB)
http://dl.dropbox.com/u/5637223/Original.jpg

iTunes AAC+ (256 ABR) (10.3 MB)
http://dl.dropbox.com/u/5637223/iTunesPlus.jpg

Nero q.7 (10.5 MB)
http://dl.dropbox.com/u/5637223/NeroQ7.jpg

Nero 256 ABR (9.92 MB)
http://dl.dropbox.com/u/5637223/Nero256ABR.jpg

You can hardly see a difference between the iTunes file and the original source.

But again, I can't hear the difference between these files, but I'd like to think that there are some people who can. And my perfectionist mind wants my music to look the same as it's source.

Which encoder do you guys prefer? I welcome your opinions, theories, and any listening tests that you can point me to.

[saratoga]

I took some screenshots of spectrums of these 4 different files-

Theres no reason to ever do this.

Which encoder do you guys prefer? I welcome your opinions, theories, and any listening tests that you can point me to.

They're both very good. Search the forums for some of the past listening tests.

[subinbar]

I took some screenshots of spectrums of these 4 different files-

Theres no reason to ever do this.

I find it very interesting to see how an encoder deals with higher frequencies. Why do you say there is no reason?

[saratoga]

I took some screenshots of spectrums of these 4 different files-

Theres no reason to ever do this.

I find it very interesting to see how an encoder deals with higher frequencies. Why do you say there is no reason?

Such high frequencies are just noise. How the encoder deals with them isn't interesting. And the spectrograph just tells you if theres a low pass in use, not if they're actually being reproduced.

[subinbar]

I took some screenshots of spectrums of these 4 different files-

Theres no reason to ever do this.

I find it very interesting to see how an encoder deals with higher frequencies. Why do you say there is no reason?

Such high frequencies are just noise. How the encoder deals with them isn't interesting. And the spectrograph just tells you if theres a low pass in use, not if they're actually being reproduced.

So can a lowpass be used, and yet those higher frequencies still exist?

[hlloyge]

So can a lowpass be used, and yet those higher frequencies still exist?

Well, it wouldn't then be lowpass filter then, now would it?
Spectrographs don't say anything about quality of encode. IF you can't hear any difference between the original and encoded file, how on earth other people who might hear it matter?
Do yourself a favor and conduct proper ABX test, to see which bitrate/setting is transparent to you on most music, than use one notch higher just to be sure.
For example, I did ABX test few years back, and found that ~128 kbit is mostly transparent to me, with exception of problem samples. So I encode my music ~160-180 kbit for safety. I am not even considering higher bit rates. I can carry around more music than before.
THAT is the right way to treat your problem, not with spectrum analyzers.

[db1989]

Such high frequencies are just noise. How the encoder deals with them isn't interesting. And the spectrograph just tells you if theres a low pass in use, not if they're actually being reproduced.

So can a lowpass be used, and yet those higher frequencies still exist?

My guess is saratoga means that the presence of high frequencies in a spectrum is not indicative of how accurately the encoder represented them versus the source.

And as several others have suggested, visual plotting is no way to evaluate the audible effects of audio processing (as indicated by #8 of the Terms of Service); as long as an encoder can fool your ears, it’s doing its job!

[Zarggg]

This doesn't get said enough (perhaps because it should be fairly obvious) but sound is sound. Our ears are designed to interpret sound in specific ways, and psychoacoustic models in lossy codecs are designed to take advantage of that. The way our ears interpret sound is vastly different from any visual representation of frequency. "Seeing" what parts of the frequency spectrum are or are not present is no better judge of encoder quality than tasting it.

This post has been edited by Zarggg: Jun 23 2011, 23:59

[subinbar]

Well, it wouldn't then be lowpass filter then, now would it?
Spectrographs don't say anything about quality of encode. IF you can't hear any difference between the original and encoded file, how on earth other people who might hear it matter?

Assuming that my ears, listening environment, and audio setup never change, then "fooling my ears" here and now would be the end of it. But the whole point of me asking is that all of these variables change. I am a musician, and I'm continually amazed at how my ears change from day to day. What sounds great to me one week (instruments/amps/etc) might need some tweaks the next.

And this is the same with music. 10 years from now I might have a very nice $5k setup and at that point I may be able to detect differences that I cannot detect now. This is why I figure I should be as thorough now as possible and seek other people's advice that possibly have a better ear and better equipment to judge lossy compression with.

Still, thanks for the advice. Like I said, I can't a/b between the two encoders unless it's a transcode. But I figured I would rather be safe than sorry later down the road if I can detect differences then.

[db1989]

What you’ve just said is, of course, completely sensible! It’s definitely a good idea to future-proof your library as far as you can. However, if you’re concerned enough, lossless compression is the only guaranteed way to do so. You may already do this, considering that in your original post you said the AACs were for mobile use; in that case, I doubt there are many big advances left to be made (through time or lofty investment!) that would reveal problems that are currently inaudible.

In any case, remember: it doesn’t matter what the audio looks like.

[greynol]

I'd like to add that the idea that better equipment will make it easier to tell identify lossy files is largely a myth, rarely (if ever) supported with objective test data by those proposing it.

Also, A/B tests lack the same control as ABX tests. As such, the results from them might not be reliable. This isn't to address any specific claims being made, but rather on general principle. Not only are A/B tests generally not double-blind, they are often conducted in such a way that they are not blind at all.

This post has been edited by greynol: Jun 26 2011, 20:46