The last year my interest in AAC has grown considerably, especially with nero releasing their encoder or free, and now I have just purchased a car stereo that plays AAC.

I was blown away with the quality of 48kb/s AAC-HE which I have been using for DJ material off a laptop, and it allows me to store almost my whole DJ music collection right on the laptop hard drive. A recent listening test has confirmed that Nero AAC is far and away the best option at this bitrate.

Unfortunately, my car stereo only supports AAC-LC, and I've discovered that Nero encodes work if using quality setting of 0.31 or greater.

With most music, this results in under 100 kb/s.

I was curious to know if there exists some kind of a chart showing the improvement in perceived quality as bitrate of AAC files encoded with nero increases. Most of the information that I see around here on AAC pegs it as being inferior in quality to MP3 and especially Vorbis at medium bitrates like this, but I suspect that things might have changed. Am I right?

Also, in general, at what nero setting would most people consider AAC to be transparent? My guess is somewhere approaching 0.5...

Thanks for any info!

I hardly have an idea to draw any objective graph to represent quality of sound.
My 2cents: q0.3 (64kbps) sounds decent while q0.325 (80kbps) has another flaws because of no SBR and not enough bitrate for LC-AAC. Generally q0.425 (128kbps) sounds neat and it's trasparent for me when I don't listen with full attention.

Where did you see this? I don't believe there is any bitrate where MP3 is superior over AAC. Vorbis and AAC are more tied as each seem to have its specific benefits but as far as I know there is no question that the newer generation codecs outperform the older MP3 (or are tied at worst).

Roughly speaking LC-AAC and Vorbis provide for the same very good quality at ~100+ kbps.
Below ~100 kbps LC-AAC is inferior to Vorbis, but HE-AAC is superior. IMO Nero's LC/HE switching q value is chosen very well.
At the widely used ~128 kbps AAC and Vorbis are transparent to most users with most music, with a bigger safety margin than mp3.
At ~160 kbps LC-AAC and Vorbis become transparent or at least not annoying even for samples that are hard to encode.

I would drop the value to below ~80 kbps in the case of Vorbis being inferior to to HE-AAC. I'd like to see some ABX tests if you can back this up otherwise.

With my own tests, I've never found HE-AAC to be superior than Vorbis below 100kbps. Only at 64kbps and below does it have a clear advantage.

Thanks for the info and opinions guys... that's about what I'm looking for.

As for the info about AAC being comparable with MP3, Vorbis, et al, this listening test seems to indicate that all the codecs are about tied at 128 kb/s or so:
http://www.listening-tests.info/mf-128-1/results.htm

Lame is almost as good as the more modern codecs, amazingly. And nero was screwed up and the results were considered irrelevant... (btw, has that bug in nero's encoder been fixed so that it can be judged fairly?)

In general I thought that nero was considered to be among the best (if not *the best* aac encoder around...

I have some (but restricted) experience with Vorbis, HE-AAC, and LC-AAC with settings of ~80 kbps and ~96 kbps. For a rough rule of thumb to me it seemed alright saying 'Below ~100 kbps LC-AAC is inferior to Vorbis, but HE-AAC is superior.'. But I guess you're both more knowledgable than I am concerning HE-AAC and Vorbis in this bitrate range so I regret having been in favor of HE-AAC.

I mean specifically bitrates around -q0 (and below) is where HE-AAC is generally superior, since it's not exactly 64. I've not tested -q0.5 or anything in between 0 and 1, but at -q1 and higher (especially with a raised lowpass), I prefer Vorbis in ABXes.

At 96kbps, I've not tested LC-AAC vs HE-AAC, but I can imagine I'd prefer LC-AAC, from what I heard in the last test with iTunes at 96 as the high anchor.

I can post some ABX logs on some normal music, if you want. Or I could do some problem samples.

If this adresses me please don't bother. No reason not to beleive you.
I'm really sorry for having said something I have too little experience with (even though it was intended to be only a rough rule of thumb).

I beleive that the bug that existed in 128 kbps test is fixed now in Nero.
You can see results for 80 kbps LC listening test now. Maybe that listening test will help you to decide which bitrate to use.

There have been a few Nero releases since that test, so AFAIK the bug in question is fixed. Personally, I've been very satisfied with my Nero AAC encodes at -q 0.4 (~125), and I'd probably use that setting for all my lossy files if foobar/Rockbox could handle the MusicIP tags (they are fine with MP3, Vorbis, and FLAC). For the moment, I've got Vorbis @ q5.

If you're into graphs, these may be worth a look: http://www.soundexpert.info/coders128.jsp

Also remember you can add the -lc option manually if you want to try really low LC bitrates.