If we detect a potential UTF-8 byte sequence we attempt decoding, otherwise we leave the bytes as they are.