Your name,
in Hangul.

It reads your spelling, not your name

There is no list of names behind this. Nothing is looked up — the tool reads the letters in the field from left to right and puts each one through a table we wrote in advance. So the same spelling always returns the same Hangul, and reopening the page does not change it. No randomness, no language model, no request to anything: the conversion begins and ends in your browser.

Before the table there is some tidying. First a Unicode compatibility decomposition, then the combining marks are dropped — which is how José is handled as Jose. But a compatibility decomposition does more than pull marks off letters: full-width Ｊｅｓｓｉｃａ, the fi ligature, circled letters and Roman numerals are all rewritten as plain a-to-z at the same step. Everything is then lower-cased, and anything that is not a letter from a to z or a space becomes a space — digits, hyphens and apostrophes vanish and leave a word break, which is why Mary-Jane comes back as two words and O'Brien as two as well.

After that, any word that produced no Hangul at all is dropped from the result. The word-by-word breakdown appears underneath only when two or more words survive that filter. So "h Kim" is two words typed and one word shown: the lone h makes nothing, and 킴 is left on its own with no breakdown.

What reaches the table is a to z and nothing else — but that is the alphabet as it stands after the decomposition above. Full-width romaji is opened out into plain letters first, so it converts exactly as the half-width spelling does, which matters if your keyboard produces full-width by default. Kana and Chinese characters, on the other hand, decompose to nothing in a to z, so every one of them turns into a space, nothing is left, and the page says it could not convert that. There are length limits too: 40 characters in the field, and a further cut at 80 inside the function. At most six words are kept, and a seventh is dropped without saying so.

Letters are regrouped into syllables

A Hangul letter is assembled from an initial consonant, a vowel, and optionally a final consonant, and this tool builds each one with that same formula. Which means a consonant with no vowel cannot be written on its own. English spelling is full of consonants in a row, so where to put them is most of what the conversion does.

The rule has two steps. A consonant with no vowel after it first tries to sit under the previous syllable as its final. If it cannot — the table gives that letter no final form, or the previous syllable already has a final, or there is no previous syllable yet — it is given a vowel of its own, always the same one, ㅡ, and becomes a syllable by itself. The table gives d, f, j and z no final form at all, so those four always take a syllable each. That is why Fred comes back as 프레드, three syllables, and Chris as 츠리스, also three: more letters out than in.

Which of the two happens depends on what follows. A consonant is only pushed down as a final when another consonant or the end of the word comes next; if a vowel follows, it starts the next syllable instead. Put Steve and Steven side by side: in the first the v drops in as a final, 스텝, two syllables; in the second a vowel follows it, so it stands at the front of a syllable, 스테벤. One exception is ours: a sound whose initial letter is ㅅ is always given its own syllable rather than pushed down, because we judged it reads better that way. That does not mean ㅅ never appears as a final — the test is on the initial letter, and the ㅅ at the end of 캇 comes from the t, which is given ㅅ as its final form.

Silent letters start with the word-final e. After a consonant it is dropped, so Kate is 캇, Mike is 믹 and Jane is 잔 — one syllable each. The rule only applies to words of three letters or more, though: the two-letter de keeps its e and comes back as 데, while dee is 디. It is also kept when the letter before it is a vowel, so Zoe stays two syllables, 조에. The pair gh is dropped, and an h is dropped unless a vowel follows it, which is why John is 존 and Sarah is 사라.

There are other ways a letter produces nothing. A fixed list of doubled consonants — ck, ll, ss, tt, pp, nn, mm, dd, ff, gg, rr — is read as a single consonant, so the second letter contributes nothing at all: Jeff and Jef are both 제프, Anna and Ana are both 아나, Matt and Mat are both 맛. The list is exactly that, a list, and bb is not on it, so Robb keeps both of its b's and comes back as 롭브 where Rob is 롭. The pair wh is likewise read as one letter, and the letter it is read as is the silent carrier, so both letters disappear: Wham is 암. A whole word can go the same way, since any word that yields no Hangul is dropped. Nothing else is treated as silent: the k in Knox survives as a syllable of its own, 크녹스.

Where it cannot be exact

The table is many-to-one. b and v are given the same consonant letter, so Ben and Ven both come back as 벤 with nothing to tell them apart. z shares with j, so Zen and Jen are both 젠; c, k and q are one letter between them, so Kate and Cate are both 캇; and th is written with the s letter, so Thomas comes back as though it began with an s, 소마스, and Beth is 베스, the same as Bes. Where two spellings land on the same Hangul, the Hangul cannot be turned back into the one you typed.

Two pairs merge in one position and separate in another, so the collision is worth stating precisely. f and p share an initial letter — Fop and Pop are both 폽 — but f is given no final form and p is given ㅂ, so at the end of a syllable they part: Leaf is 리프 and Leap is 립, Olaf is 올라프 and Olap is 올랍. l and r share their pair of letters, which is why Lee and Ree are both 리 and Rita and Lita are both 리타. But an l with a vowel before it and a vowel or a y after it is doubled into a final ㄹ plus an initial ㄹ, and an r in the same place never is: Kelly is 켈리 and Kerry is 케리, Alan is 알란 and Aran is 아란. Neither pair is a blanket merge, and one example of either is not evidence about the other position.

The other half of the same point: some letters never come out at all. Nineteen consonants can begin a syllable and this table uses fourteen; the five written as a doubled consonant — ㄲ, ㄸ, ㅃ, ㅆ, ㅉ — never appear, whatever you type. Of the twenty-one vowel letters only fifteen are ever produced: ㅒ, ㅓ, ㅕ, ㅙ, ㅚ and ㅢ do not come out. Of the twenty-seven possible finals only seven do, ㄱ, ㄴ, ㄹ, ㅁ, ㅂ, ㅅ and ㅇ. That was checked by running every one-, two- and three-letter string plus four hundred thousand random longer ones through the function: the rest never appeared once. What this tool can write is a subset of what Hangul can write.

And because the table was written around English spelling, romanized names from elsewhere are read by English rules. In Ayaka the leading ay matches the rule for a single vowel, so the y sound never gets a syllable and the result is 애아카. Ryo gains a syllable because no vowel follows the r, leaving it to be carried by one of its own, 르요, and Tsubasa splits for the same reason into four, 트수바사. The breaks can land somewhere other than where the original language would put them.

It should also be said that the tool holds no list of real names. Feed it Kim, Park or Choi — names that were Korean before they were romanized — and it does not consult what they were written with. It re-reads them as Latin letters and hands back 킴, 팔크, 쵸이: the sound of that spelling, not the name it came from.

How far this goes

This is not an official romanization run backwards, and it is not a translation of what a name means. We wrote the table, and it could have been written differently. Since several Latin letters land on one Hangul letter, no spelling can be derived from another with a single right answer waiting at the end. What appears here is one way of writing it, produced mechanically by rules that are all on this page.

The name you type is never used for a chart. No birth date and no birth time are asked for, and this page does not load any part of the chart engine. Pressing the button runs the conversion described above and puts the result on screen, and that is all. The link to the saju reading below the result carries nothing with it either — that page asks for your birth date from scratch.

As for what it is good for: a label in a notebook, something to practise on early in learning the language, a line to show someone. It is not built to decide the name on a document or the spelling on a business card. If a Korean speaker who knows you writes your name differently, use theirs. What this page offers is one suggestion with all of its workings shown.