I still don't see the point, but here's a simpler and more canonical way. It shows how to do it for both an encoded and decoded string.
#!/usr/bin/perl
use warnings;
use strict;
use Text::Unaccent::PurePerl qw(unac_string);
use Encode qw(encode);
binmode *STDOUT, 'encoding(UTF-8)';
my $string = "c\N{LATIN SMALL LETTER O WITH CIRCUMFLEX}te";
my $rstring = encode('UTF-8', $string);
print "$string $rstring\n";
my $direct = unac_string($string);
my $via_bytes = unac_string('UTF-8', $rstring);
print "$direct $via_bytes\n"; # cote cote
map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]
| [reply]
[d/l]
[select] |