What's amusing is that speakers may have been optimizing themselves for LAME, since it's pretty much the dominant encoder.