Characters showed up as Chinese when the raw data was interpreted directly as UTF16 (both little and big endian)