This is an ironic question, considering how all the best media databases on the web originally started as volunteer efforts by aficionados. Media-metadata gathering is possibly the best case-study of effective crowdsourcing.

So the answer is basically “just empower the community to do it”.