There may have been discussion on this already, but I did not spot it on an initial russle through.

I have a 30G iPod and am wondering what the best AAC encoder is? Do they have varying charactaristics? Is one likely to improve faster than others? What are the free ones? Are they any good? Is it worth paying for Nero? Do the AAC files have to be in an MP4 container? Is their likely to be a favorite tagging mechanism for MP4, if so what is it and is it compatible with the iPod (I read something about fb2k being able to write some compatible tags in .7 b9)

Lots of stuff here, but this seems more tricky that just downloading the latest copy of LAME

http://rarewares.hydrogenaudio.org/test/aa...st/results.html

Thanks for that, v.useful.

Just wondering though, as there is no date on this test, when it was done?

Would I be better off buying QTPro or waiting for Nero6 on a windows platform (as Nero6 is new, is the encoder likely to have improved much)?

From reading some posts it seemed like a few people worked on the codecs and thus I though they might have the inside line, as in, what do you use?

About a month ago. You'd have to follow the link near the top to find the date (along with instructions) for the test.

I'd be more interested in comparing average bitrates at the point in which the encoders become transparent.

AAC files have to be in a mp4 container.

For tagging it only matters that which ever program is used to upload the mp4 files to the ipod is able to read the tags, the ipod its self reads no tags. Tagging has just stabilized for mp4 making other programs, such as foobar, dbpoweramp compatible with the tags written by iTunes, although I have to say it seems Apple have different tagging between iTunes and QuickTime...

Hrm. I see a problem in your request. You see, "the point in which the encoders become transparent" varies from person to person. My personal transparency bitrate will surely be much lower than, say, Guruboolez's one (my hearing is crappy, his is amazing)

Well, that kind of precision would be nice - I don't have very good ears either. But I guess I really mean "the point at which the encoders get LAME -aps transparency".

Well, in this case it won't really be a listening test.

You see, to test "transparent bitrates", I would have to use only problem cases, and only the real golden eared members of this community would be able to participate. Therefore, I probably wouldn't be able to draw any significant result from the test. I mean, the samples would be transparent for most people anyway!

I think a better question is what are the recommended settings for AAC? Lame has its -aps, but for AAC there is no clear answer. QuickTime and Sorenson both do well, but they're not variable bitrate and Sorenson is limited to 128 kbps (in the trial version, at least), and if there's one good thing r3mix taught me, it's the importance of vbr. I originally decided to use Nero with Normal quality, until spoon said in a post that he recommends extreme. As a result, I'm a little confused about what I should use...


We have some common listening tastes, but may I ask what your gear is?

Edit: linked to spoon's post.

The full version is also limited to 128. And also CBR only.

I have very strong reasons to believe VBR will be implemented in QuickTime AAC soon.

For Psytel/Ahead AAC, I always recommend the -normal preset. Might not be transparent on all cases, but still it's enough (for me at least) for non-critical listening environments. If you want the warm-fuzzy-overkill-feeling ™, go with -extreme.



Well, well, did Spoon conduce listening tests comparing normal versus extreme?



Crappy ears, crappy Philips headphones. Not a paradigm for anyone...

Yes, but would you recommend QuickTime 160 kbps over Psytel/Ahead AAC -normal? If not, why do you use it?

Dude, I don't recommend anything. All I recommend is that people test by themselves and find what suits them better. :B

That said, I _believe_ QuickTime 160 is on par with Psytel/Ahead -normal. personally, I use Psytel, mostly for batch encodings and because there's no QT VBR mode. Once (if ever) QuickTime starts offering a VBR mode, it will become the encoder to beat quality-wise.

I've read enough posts to know that's what you recommend by now . However, results have been inconclusive. I tried ABXing guruboolez's Bachpsichord.flac sample with quicktime all the way down to 112 kbps, and failed all the tests . Usually I do a reasonable job, but I don't know what happened there , so that's why I asked what you recommend. Thanks for the tips!

It seems that I'm more sensitive to harpsichord nuances than some others people on this board. On the other side, I sometimes feel lost with some others samples, like electronic one. I like this instrument, often listen to it, a small difference is often giving me a bad feeling (you know, the famous warm fuzzy feeling™) that I can usually ABX at moderate bitrate, and with a lot efforts at higher one.

Recently, during Roberto's public listening test, Garf - another member - gave a very strong notation to a totally different kind of sample (metal - can't remember the name). Personnaly, if I could ABX the sample, I don't notice hard difference, and wasn't annoyed by them).

Therefore, there are great disparity in individual perception. Not only a question of training : musical tastes are certainly playing a consequent role, here.

No, but I will say this in my defence - it is horses for courses, extreme gives about 192Kbps, now if you had a flash player then I wouldn't recommend extreme, but on a X Gig iPod then extreme is a good bet, I am guessing it is a no brainer for 'problem samples' you might happen across.

hmmm, I can't ABX the difference on a latin jazz mp4 made with nero 5.5 on normal vrs. extreme. That's with a Stereo-link 1200 and Shure e2c's which should give me pretty repsectable quality.

I've heard people say electronic music is generally easier to ABX? I'm not a huge Fatboy Slim fan -- would something like Tricky or Massive Attack be a better choice?

What I'm getting at is that if mp4's at normal are transparent for me, I'd rather use that than mp3's since they are a bit smaller. I'm not interesting in testing problem samples, just music that I listen too.