I have a very large and multilingual collection of East European music (numerous alphabets, diacritics, etc). My problem concerns the finding and deletion of duplicates. The music is archived as follows.

The overwhelming number of recordings are EPs, LPs, etc—not individual “singles” or lone MP3s lacking artwork. This means that almost every release is in a folder, containing both the MP3s/FLACs/WAVs, whatever - and the artwork as a JPG or PNG file. Those multiple folders were once collected on individual hard drives, which in turn were then transferred to a single multi-rack NAS server (offline), which is where they currently sit.

So the entire collection (2M+ files) consists of maybe 30 folders, each representing an older external HD, and each of those HD folders contains very many sub folders (each = an LP, EP, etc,...). 

Perhaps not surprisingly, we have two problems. (1) DUPLICATE files and LPs, which used to be on separate external HDs but are now in one location. (2) The SPEED searching for duplicates across multiple racks (on an aging NAS sever) is painful. Time for a major backup, I know. 

As I decide whether to upload everything to a cloud-based DB or invest in a new server, my question to the forum is this: 

**Which software and tactic do you trust most to find/delete duplicate digital tracks? I’ve had partial luck with some. (1) Gemini2 is interesting but freezes when dealing with even mid-size tasks. The online support is awful, too. (2) SongKong is good and well supported, if not old-fashioned, but I worry greatly about any program, for example, confusing LPs and compilations, say—and merely punching permanent holes in compilations, deleting what it considers to be “copies” of LP tracks. 

Given that my collection is so big, I’d love to rely on some form of automation, if possible. 

Any suggestions, please? 
Thanks! David