use v5.12;
use warnings;
use Devel::Peek;
use Data::Dump qw/pp/;
use JSON::XS;
use utf8;

my $str = "\x{feff}"
  . '["whät","över"]';                  # Internal Unicode b/c of use utf8

Dump($str);                             # hence shows UTF8 flag

$str =~ s/^\x{feff}//;                  # strip BOM by Unicode code-point

Dump($str);                             # shows UTF8 flag

my $JSON = JSON::XS->new;               # coder for all unicode in/out

my $data = $JSON->decode($str);

warn pp '$data: ', $data;               # ["wh\xE4t", "\xF6ver"]
                                        # NB: \xE4, \xF6 correct codepoints for umlauts 
                                        # even if each character needed 2 bytes

my $str2 = $JSON->encode($data);        # roundtrip

Dump($str2);                            # shows UTF8 flag

warn '$str eq $str2: ', $str eq $str2;  # same