Slashdot Mirror
Hydrogenaudio AAC Listening Test Results
caffeine_monkey writes "Hydrogenaudio's AAC public listening test, previously posted on Slashdot, is now over and the results are in. The test compared five codecs at 128 kbps, including Psytel, Nero, Sorenson Squeeze, QuickTime, and FAAC. The winner? 'QuickTime is a clear winner, performing much better than the competition. Sorenson Squeeze, Psytel AACenc and Nero are tied, with Sorenson slightly higher than the others. Faac is clearly the worst.'"
...because it was an AAC test, not a general compressed audio test.
Would I be correct in assuming that it is the QuickTime codec that is used in iTunes? I can't imagine Apple would go to the trouble of writing two different codecs.
"Reality is merely an illusion, albeit a very persistent one " -Albert Einstein
What I'm curious about is, there was some discussion before about the differences between the original AAC encoder that came with QT, and the newer one that now ships with QT 6.3 (and ties to iTunes). The original encoder was said to have sucked. This one, if I'm reading this right, is now very good...?
Anyways, I must have lead ears. I used to rip my MP3s at 160kbps, now I do 160kbit AACs, and cannot really tell the difference. AAC seems a tiny bit better maybe but could be a placebo effect.
If Jesus wants me it knows where to find me.
:) now that's amusing.
seriously, audiophiles are indeed full of s&*t. sorry to say, but when they have to listen and compare stuff, they are very inconsistent.
mostly audiophile gear is a jerk-off session of wannabee music expert wealthy bast^H^H^H^H guys who can afford to spend large anounts of money on equipment
what also strikes me as interesting is that most musicians i know don't give a shit about their hifi set. as long as they can get the general idea of what a song sounds like and what the different instruments do. or something.
anyway, i'm getting way-off-topic here, so i better stop
" Its pretty widely known that WMA offers superior quality"
Actually, the tests that show this were at the 64kbs range which is not useful except for voice. The break-even point is 128, and above that sampling rate, virtually everything (including MP3) is superior to WMA.
So I guess if you want a good voice-streaming codec, WMA is your codec. But for music, it isn't very good.
But perhaps you're a person who primarily listens to books on tape or something like that?
I'd like to see some psychophysical analyses of these things. Failing that, see what blind test/retest data are like. I do a get a bit puzzled when audiophiles claim to be able hear differences between things outside the ability of people tested with very high level equipment in laboratory conditions to discern or discriminate between. I'm not saying they are necessarily lying or imagining things, but on the other hand at the present time some (not all) claims made by audiophiles are akin to those made by people claiming to be psychic.
Plays violent online games as: Nerfherder76
Only problem I see is: in this fast evolving softwareworld, if they sequentially test encoders for different formats, the first one tested has a really good chance to be less than state-of-the-art, by the time they finish testing. So when it comes to a head to head comparison of the formats themselves, it is possible that in the mean time there were better encoders developed for (in this instance quictime,) giving it a disatvantage. I hope they keep an eye open for new releases.
Studio technicians and musicians aren't generally known for having 'golden ears'. In fact long hours in front of loud monitors and guitar amps tends to make for poor hearing of very high frequencies.
Most people agree that Vorbis (often referred to as Ogg, but that's the container) is superior at low bitrates as its artifacts are usually less offensive than those of low bitrate MP3s etc. Vorbis is certainly NOT the best lossy codec at high bitrates however. I would be curious to know when this test was done, and what MP3, AAC and MPC encoders were used.
The statement that at above 128 kilobits/second none of the codecs were clearly distinguishable from CD is laughable and leads me to believe the testers either have 'tin ears' or are untrained in spotting encoder artifacts. Training really matters here! It makes a big difference when you know how an encoder screws up a signal, and how to spot it.
The general consensus among those 'in the know' is that in order to reach transparency or 'near CD quality' with MP3 you'll need a good encoder like LAME and a VBR bitrate in the neighborhood of 220+ kbps. (Use the standard preset.)
I wonder if these tests were conducted using loudspeakers. Most codec testing is done with headphones which tend to be better for spotting tiny differences in encoded material.
So your basic complaint is that you don't know the definitions of the scales, so you're assuming that they might as well be random. Go read http://ff123.net/64test/practice.html (which is where they explained the testing process to the volunteers).
Their scale, btw, is:
5.0 = Imperceptible
4.0 = Perceptible, but not annoying
3.0 = Slightly annoying
2.0 = Annoying
1.0 = Very annoying
Yes, I agree that people will vary in their interpretation of the terms (what's the difference between "annoying" and "very annoying"?), or have other individual aberrations (rank all encodings of the song as great if they like the song) -- but in a large sample size, individual variation is compensated for (which is one reason why sample size matters).
If you couldn't apply statistical techniques to subjective data, opinion polling wouldn't work, and of course it works quite well. You just have to know that you're dealing with subjective data and treat the results accordingly. Rather obviously, a codec rated 4.0 is not "twice as good" as a codec rated 2.0, for example. But it's still valuable to know that with a 98% certainty, people prefer the output of the first codec to the second.
If you want absolute certainty, instead of percentages and approximation, you shouldn't be doing statistics, you should be doing
Enable 3D printed prosthetics!