[gzeus]

I don't understand why iTunes doesn't have a vbr aac option for encoding. Vbr mp3 with lame gives better results than cbr aac, which is the only option iTunes gives you. I have to go through the less than satisfactory process of ripping with EAC and encoding with Nero on a windows machine to get vbr aac. So I was wondering, is the impending release of Tiger mean that iTunes will also be updated with the vbr option?

[Turias]

I'm not exactly sure what the details are, but I thought I heard that AAC is pseudo-VBR. Something like it's not VBR by definition, but it acts (and works) about as good as true VBR.

Can anyone elaborate for me?

[chabig]

I think the expectation is the Quicktime 7 in Tiger will include support for the creation of VBR AAC.

Chris

[bmedina]

Even with the limitation of CBR, iTunes still has the (arguably) best AAC encoder publicly available.

And VBR doesn't always give better results than CBR. iTunes' MP3 encoder is proof of that. Hopefully Apple is using this time to make a good implementation of AAC VBR for Quicktime 7. If the CBR version is anything to go by, the VBR version should be amazing.

[Busemann]

Originally posted by gzeus:
Vbr mp3 with lame gives better results than cbr aac

Not true according to blind-tests:



As you can see, iTunes AAC is slightly better than LAME VBR at 128kbps. The iTunes encoder has also been improved a great deal since this test was done.

The current AAC implementation uses bit-reservoir, so the bit-rate can vary somewhat depending on the complexity of the music (if you encode digital silence, you'll see the resulting file is just a few kb's).

QuickTime 7 (and the next iTunes) is expected to include a true VBR mode, though.

[gzeus]

In all deference to the 18 people in the study (their atrac3 rating helps with their credibility), lame with the -extreme flag has sounded better to me than cbr aac. Every encoder has its sweet spot, and anecdotal reports suggest aac's is around 128 (the listening tests I've seen generally equate 192 mp3 to 128 aac). I'm really curious how well the new itunes vbr encoder extends out in the higher kbps ranges as I've been using the -audiophile option in Nero (sue me, I'm a snob). I guess I'll wait until Tiger is released before encoding the rest of my cds.

[bmedina]

Originally posted by gzeus:
In all deference to the 18 people in the study (their atrac3 rating helps with their credibility), lame with the -extreme flag has sounded better to me than cbr aac.

Yes, and do you know what kind of bitrate preset extreme entails (hint: it's CBR)? What bitrate were you using with AAC in the comparison? And what software did you use to compare the two?

I've been using the -audiophile option in Nero (sue me, I'm a snob).

If you're so concerned, why not just use a lossless codec? It'll be roughly double the bitrate you're using now, but you'll get peace of mind.

if you encode digital silence, you'll see the resulting file is just a few kb's

When did you try this? I know that earlier (around iTunes 4.2), the file would be 128kbps. The resevoir is very small, so even silence should eventually be encoded at 128kbps. Perhaps Apple thought that was silly, so they coded an exception for pure silence.

[gzeus]

Yes, and do you know what kind of bitrate preset extreme entails (hint: it's CBR)?

The extreme preset is vbr . You might be confusing it with the insane preset.

What bitrate were you using with AAC in the comparison? And what software did you use to compare the two?

192kbps aac. Subjective listening test, same type of test as what Busemann quoted.

If you're so concerned, why not just use a lossless codec?

Lossless just kills the battery life on the iPod so it wasn't an option. Besides, aac is also supported by Philips and Panasonic portables, and more companies will be signing on when people start recognizing aac as mp3's successor. Encoding in Apple Lossless would limit me to the iPod, just as encoding in Flac would limit me to the Rio Karma.

[Busemann]

Originally posted by gzeus:
192kbps aac. Subjective listening test, same type of test as what Busemann quoted.

Nope, I quoted a proper ABX blind-test. I suggest you do the same kind of test before you make up your mind