Using PHash Hashing Scheme to seamlessly identify user media

Im new to _beets_, I really liked the idea and it works great. But if someone has a huge unorganized library then it becomes a pain to answer questions and he may just give up. And human intervention may not be 100% correct, I ended up submitting a lot of incorrect choices while trying to import my music database.

My thoughts are aligned with a hashing scheme. The client will calculate hash of a particular file and send it across to server which is a repository of **{HASH -> METADATA}** key value. Client will get the most accurate result possible, and can then organize the media in an appropriate way. This thing works brilliantly in DC software (Linux DC++, ApexDC++, etc). Also shazam uses a similar technique to identify music.

No file, test, or entry point is named. Start by mapping beets' music import flow and how it currently handles user questions, then investigate the proposed client-side hash and server-side HASH-to-METADATA repository. Done would require a defined hashing scheme, service boundary, metadata behavior, and an implementation plan.

Written by the indexing model from the issue text.