10x sill ! again, i believe i know what i'm doing. but you surely have the freedom to do something else. imho, the combination of azid's dialog normalization + static pregain of 10db + dynanmic postgain plus + disabling scaling of alt-preset to 0.93 by using '--scale 1' is the right way to go. why won't you compose a small comparision test and see which method ends up with a louder mp3 :) ? btw, where's hongkongfu ?