I would recommend a lookup table, rather than repeatedly doing the work of de-interleaving. De-interleave every possible 2-byte input into two characters once, then read your data as 2-byte shorts and lookup the output characters.

use strict;
use warnings;

my @masks = map 2**$_, 0..15;

sub deinterlace {
  my $data = shift;
  my ($odd, $even) = (0, 0);
  $odd  |= ($data & $masks[$_*2  ]) >> $_ for (0..7);
  $even |= ($data & $masks[$_*2+1]) >> ($_+1) for (0..7);
  [pack('C', $odd), pack('C', $even)];
}

# Build lookup table
my @lut;
$lut[$_] = deinterlace($_) for (0..0xffff);

# Example data
my @d = unpack('S*', 'Some data to de-interlace.'. "\xfa" x 6 . "\0" x
+ 6);
printf "%04x ", $_ for @d;
print "\n";

# Building output strings
my ($odd, $even) = ('', '');
for (@d) {
  my @bytepair = @{$lut[$_]};
  $odd .= $bytepair[0];
  $even .= $bytepair[1];
}

printf "Lengths: %d and %d\n", length($odd), length($even);
printf "%02x ", $_ for unpack('C*', $odd);
print "\n---\n";
printf "%02x ", $_ for unpack('C*', $even);
print "\n";
[download]