I'm searching a short way to uniquely identify any given track in my music collection by a short, unique and alphanumeric ID [A-Za-z0-9-]. (Reason: see signature.)
Ideally, I would like it to be human readable, or at least interpretable, so that a track could at be recognised among a short sequence of those IDs. I have come up with the following title formatting string that I use with foo_dynfil to calculate a field %_uid%:
$replace(
$if2(%release%,$lower($abbr($ascii(%album artist% %album%),6)))
$right(%date%,2)
$lower($left($abbr($ascii(%title%)),2))
$ifgreater(%totaldiscs%,1,%discnumber%,)%tracknumber%
,-,,?,,'',,",,/,,\,,!,,&,,:,,.,,+,,'(',,')',, ,,@,,'[',,']',,~,,#,,%,,_,,^,,|,,*,,$,)
On my 25000 track database it creates one collision, with an average length of 13.5 characters per song. Has anyone here done something similar already which is a) shorter and/or b) less likely to have collisions?