I recommend using a "canned" regex to extract all http url's and then use grep to select the .mp3's.

use strict;
use warnings;
use Regexp::Common qw /URI/;
my $string = "xxx http://perlmonks.com yyy http://foo.mp3 zzz";
$_ = $string;
my @urls = grep {/\.mp3$/}  /($RE{URI}{HTTP})/g;
local $" = "\n";
print "@urls\n";


OutPUT:
http://foo.mp3
[download]

This approach should find all valid http url's with no false matches while saving your hair.