AMTuring
17th April 2007, 06:06
Just a short post. I have finally released my magnum opus in the codec comparison field. The program is available for download at http://amturing.batcave.net.
When applied to Sebastian's codec comparison it shows that the codec rankings are Ogg Vorbis, iTunes AAC, WMA, Nero AAC, and LAME mp3 bringing up the rear. This is in complete agreement with the listening test (128 kbps settings) Doom9 discussion of Sebastian's Listening Test (http://forum.doom9.org/showthread.php?t=107393).
Using 2000 analysis channels the correlation with the human listening test is 95%. That is good enought for me.
This program is based on a model of the human ear using oscillators (scilia in the ear) with frequency dependent damping, logarithmic spacing, and A-weighting to incorporate all of the elements of psycho acoustic models.
I have been sitting on this for a while. My intention is to include a compete paper describing the results, but I have been lazy.
When applied to Sebastian's codec comparison it shows that the codec rankings are Ogg Vorbis, iTunes AAC, WMA, Nero AAC, and LAME mp3 bringing up the rear. This is in complete agreement with the listening test (128 kbps settings) Doom9 discussion of Sebastian's Listening Test (http://forum.doom9.org/showthread.php?t=107393).
Using 2000 analysis channels the correlation with the human listening test is 95%. That is good enought for me.
This program is based on a model of the human ear using oscillators (scilia in the ear) with frequency dependent damping, logarithmic spacing, and A-weighting to incorporate all of the elements of psycho acoustic models.
I have been sitting on this for a while. My intention is to include a compete paper describing the results, but I have been lazy.