I can see two possibilities off the top of my head, but it would be very helpful if you would more fully describe the "files" that are downloaded where mp3s were expected.

Are these files the same lengrh as the expected mp3? If so, try renaming the file to an *.MP3 because for some reason your servent didn't...

More likely, are these files disgustingly small. Are they .URLs masquerading as mp3 QueryHits? These are spam, and I've only seen very few of them to date... What happens is you Query, and some spammer returns his spam as a queryhit. When you request the file (which appears as a reasonable-length mp3), you send your GET request and are rewarded with a 122 or so byte file that holds only a URL. *******s!

So what does your non-working file look like? What's its extension and size?