If you have a newly ripped album with multiple tracks with same name and length (e.g. remixes) then all files are allocated to the first track with the name rather than attempting to use track number or attempting to split the files up between the tracks.

This is because album.py#match_tracks does optimisation on a file by file basis, rather than all files at once.

It might be better to e.g. get all the match values for each file into a sorted array, then go through these in decreasing sequence, allocating files to tracks that are unallocated.

This would hopefully put each file against its best match.

Note: Some tweaks to the weighting algorithm might be needed to e.g. increase the weightings for track number and length and to change the sensitivity curve of the title match so that small differences like the mix at the end have the best effect on allocation and avoid similar tracks being allocated the wrong way around.