HTTrack, doesn't Do directory structures.  It does Files-By-Link. It then
creates any directory required by the files.

So you will need to grab the HTML, to get the links, to know where the files
are, to get a directory structure. Luckly HTML files are tiny.

But you also want the MP3 file locations (directorys) which might not have a
html file in it.

So you'll need to grab the MP3 files too (which is what you're trying to
avoid). So on your MP3 filter use the broken syntax "*[<NN>MM]"  set to
something like *[2<N>1] This should cancel the MP3 Download at 2KB.

This will leave you with the Directory Structure + html-files +
tiny-broken-mp3-files

It then easy to windows search function with the value *.*
Sort by Type, select all the files (not folders) and hit DELETE

You'll be left with a empty directory tree