++ for taking unpack and making it nearly readable :-). However, I have to wonder why, if you're writing it anyway, you're joining on spaces, and then splitting on spaces. Wouldn't it be slightly more readable if you did:

my $h = unpack2hash(
  [
  'l:$songid',
  'c8:@signature',
  'l:$genre',
  'f:$bpm',
  's3:@level',
  's:$unk',
  'l12:@unk2',
  'a24:$genres',
  'l2:@unk3',
  'a32:$title',
  'a32:$subtitle',
  'a32:$artist',
  'a32:$noter',
  'a32:$musicfile',
  'l:$jpg',
  'l3:@unk4',
  'l4:@notepos'
  ], $data);
[download]

This would arguably be faster (arguable because even going down that road might be a premature optimisation), but I also just think it cleaner. Of course, if you're doing this anyway, you could also just pre-split a bunch of the text, e.g.:

my $h = unpack2hash(
  [
    [ qw(l $ songid) ],
    [ qw(c8 @ signature) ],
    #...
[download]

Also think about reversing the order of the parameters, then we could get rid of the array-ref:

my $h = unpack2hash( $data,
    'l:$songid',
    'c8:@signature',
    #...
);
[download]

These are all mere suggestions, so take, or not, as you want - which is why I didn't repeat the previous suggestion to possibly pre-split. Of course, you could also be dynamic and if passed a scalar, split it yourself, but if you're passed an array ref, you could treat it as pre-split. But that would depend on usage.