diff --git a/t/cmd/tokenizer.pl b/t/cmd/tokenizer.pl
index bf948e8..4e8b2d0 100644
--- a/t/cmd/tokenizer.pl
+++ b/t/cmd/tokenizer.pl
@@ -2,6 +2,7 @@
 use strict;
 use warnings;
 use FindBin;
+use Encode;
 BEGIN {
   unshift @INC, "$FindBin::Bin/../../lib";
 };
@@ -14,7 +15,7 @@
 
 # Read lines from input and return boundaries
 while (!eof(STDIN)) {
-  my $line = <>;
+  my $line = decode_utf8(<>);
   for my $text (split(/\n?\x{04}\n?/, $line)) {
     $tok->tokenize($text);
     print join(' ', $tok->boundaries), "\n";