Roberto Amorim of RareWares has published the results of the public MP3 at 128kbps listening test with interesting results. As we can see, the latest LAME 3.95 achieves the best quality, which is yet another great achievement by this open source project. GoGo is a speed optimized variant of LAME and is clearly worse than the original. Xing performs much better than expected.
Worthless exercise. The graph is bogus as it the y-axis intercept is not zero. It if was you could see that the difference between the codecs is insignificant. A normal person will not be able to hear the difference.
Bmorey, That's some BS that you are saying. The test was done by 'normal people' and differences were heard. Read some discussions about the test: http://www.hydrogenaudio.org/index.php?&showtopic=18378
I would have to aggree with bmorey. The difference of 0.7 is really not that significant for only 12 samples. Even though "normal" people were tested, what ever that means. This to me says that people really can't tell the difference.
Testing codecs at a fixed 128bps bit rate that have been specifically developed and optimized for VBR use seems kind of useless to me (all of them except for FhG and maybe Gogo as far as I know). I would say the test results might actually be misleading as to the quality one can expect from a given codec when it is used the way it was meant to be. How about testing a single codec in the modes it was designed for working from the largest most inefficient file size down to the point where "normal" people start to hear artifacts? That would be useful information for me particularly if the resulting file size was included. I've thought about sending e-mails to RareWares about this before but I hate to dis all their hard work. It's not like I've helped out. Well, even with the "rigged" test, LAME is still the best. Once you've tried APS you'll never go back... And it's FREE!!!
