Funny you should say that. I've always thought it looked like a mouth, too, and at a LAN party once, someone complained about that "thing that was making faces at them".
I have had a related idea for a visual that could perhaps be implemented as a WinAmp plug-in, or maybe even as an Empeg visual (are you reading, Toby?)...
It harks back to my childhood, when a local PBS station had some sort of classical music concert programming. I don't remember the program's name. But I remember that durnig the end credits, they would have the faces of two mimes on the screen. They would play a classical recording, and these mimes would "mouth" the sounds of the orchestra, each one taking a group of instruments. For instance, one would mouth the "pom pom pom" of the basses, and the other would mouth the high melody parts. It's kind of hard to describe, but it was hilarious to watch in a cool/creepy sort of way.
I figured you could do that pretty easily in software, with beat detection and such. Split up the frequency bands into 2-4 "mouths", create a few bitmaps representing the frames of animation, and you're set.
___________
Tony Fabris