On Tue, 2006-03-14 at 01:55 -0800, Benton Lam wrote: > One is an option to all the duplicates in the library, and the other is > that for each song entry, showing the song type (i.e. encoder type), or > even just the file extension is great. The second of those is the last part of bug 167659, and has some patches attached to the bug. The first (duplicate detection) is more difficult, as detecting when one file is a duplicate of another is a hard problem. Some potential solutions are: * MD5 (or similar) summing of files. This will only detect *exact* duplicates, and not work if anything changes the file, such as editing tags. * Matching <duration, mime-type, artist/album/track>. Would work reasonably well - but not detect badly-tagged tracks, duplicates in different formats, or copies with extra leading/trailing silence. * Audio fingerprinting (e.g. musicbrainz). Fairly slow, but would work well. Also useful for implementing tag-lookup on audio files. Cheers, James "Doc" Livingston -- "I had my first real beer (real meaning not sneaking it, or the .5 crap), and amazingly enough, I wasn't immediately surrounded by girls in bikinis. Go figure...." -- Rob Hoadley
Attachment:
signature.asc
Description: This is a digitally signed message part