Chinese Name Romanization Converter 2026

Readings come from the McBopomofo input method’s open data (21,745 characters). Where a character has several readings, the default is the Taiwan reading given in the Unicode Unihan database. Fifteen surnames with a special reading (such as 曾, 沈, 單 and 解) are set by hand.

What this tool cannot do

1. It shows how the table would spell a name. The spelling on a real passport depends on the holder’s previous passport, the family’s surname spelling and the holder’s own choice; the Bureau of Consular Affairs does not impose a system.
2. For characters with several readings it can only list them for you to choose. About 360 rare characters have no frequency data, so the default is simply the first in the list.
3. Seven syllables in the reading table are not in the official table. The tool flags them, and those spellings are only indicative.
4. A spouse’s surname added in front, or a double surname, cannot be detected from the characters. Use the surname-length switch.
5. Indigenous-language names do not go through this tool. For someone registered in an Indigenous script only, that spelling is the passport name.
6. Taiwanese and Hakka readings cover the thirty most common surnames only. For given names use the Ministry of Education dictionaries.
7. The reverse look-up does not know tones and cannot tell homophones apart.
8. Speech uses the voice installed on your device. Without a Taiwan Mandarin voice the button does not appear, and accents differ between devices.

The official table in full

Bureau of Consular Affairs, Passport Name Romanization Table, edition of 29 May 2026. The original is marked “for reference only”.