MP3 vs AAC vs OGG
February 16th, 2006 by George TyshchenkoWhile OGG format seems to provide an extended frequency response it sacrifices linearity which is important for natural reproduction. AAC format seems to roll off at around 17 kHz but provides an incredible flat response which gives this format great advantage in certain cases. The files used in this test were 128 kbps.
| Test | MP3 | AAC | OGG |
| Frequency response (from 40 Hz to 15 kHz), dB: | +0.32, -1.83 | +0.10, -0.21 | +0.21, -0.19 |
| Noise level, dB (A): | -96.8 | -97.9 | -101.2 |
| Dynamic range, dB (A): | 96.1 | 96.1 | 93.5 |
| THD, %: | 0.0009 | 0.0004 | 0.0013 |
| IMD + Noise, %: | 0.020 | 0.034 | 0.468 |
| Stereo crosstalk, dB: | -98.4 | -98.2 | -101.7 |
You can view the full RightMark report here
August 23rd, 2006 at 10:28 am
This is a very interesting test. My only issue with this is the lack of detail regarding encoding information. You don’t say which encoders you have used for each codec.
For example, LAME is by far the best (in listening tests) MP3 encoder, likewise with the Nero AAC encoder. Perhaps publishing the results of multiple encoders, or at least publishing the names and versions on the encoders that have been used? I expect this information would have a considerable effect on the outcome.
Otherwise, the 15kHz roll off of AAC worries me and makes me wonder if this is why my iPod seems to lack treble. Perhaps MP3 is a better choice after all :S. Can this sort of thing be corrected with preemphises?
October 14th, 2006 at 2:44 pm
This sort of test, for audio compression formats, isn’t very useful. Plots of frequency response mean nothing if you cannot hear the differences in real situations. Blind listening tests are much more useful.
Also, lack of ‘linearity’ to the order of +/- 0.2 dB would certainly not be audible across the frequency spectrum.