But the ideal solution would be to have some way of generating fingerprints for each audiobook, and then build up a database which matches that fingerprint to the correct metadata. That way the work of organizing and tagging large collections could be crowdsourced; this is what other communities have done.

I was referring to this bit, which is exactly what the Musicbrainz database does for music albums.