[pavel_lishin]

Well, if I were to do it, I'd probably implement a few checks.

I would check to make sure that the ID3 information was no more than, say, 10% of the total file size. (I'd probably adjust the percentage based on actually inspecting a large collection of files; say, the ones available for download through Amazon.)

I'd probably make sure that the resulting sound output had a regular pattern - a steady beat, etc - and flag for review anything that wasn't, probably with a rating of how likely it is for the file to not be an mp3.

And obviously, scan the file to make sure that it doesn't obviously contain an archive, images, ASCII text that falls within the KJV bible distribution, and check for other formats; someone renaming stolen_e_book.pdf into music.mp3 would probably be pretty obvious.