Skip to content
JTK

Romaji to Hiragana Converter

Type Japanese in the Latin alphabet and get hiragana back instantly, in your browser.

Hepburn or Kunrei both work. Converts as you type.

Type romaji into the box above and the hiragana appears as you type. Nothing is uploaded and nothing is installed — the conversion runs in your browser, so it works offline once the page has loaded.

What romaji is

Romaji is Japanese written in the Latin alphabet. It is how Japanese is typed on a standard keyboard, how station names appear on signs for visitors, and usually how a learner meets the language before learning to read it.

It is a transcription of sound, not a script Japanese is actually written in. That matters here, because more than one romanisation system exists. Modified Hepburn spells the sounds as an English speaker would read them — shi, chi, tsu, fu, ji — and is what you see on signage. Kunrei-shiki is more regular and is taught in Japanese schools — si, ti, tu, hu, zi. Both describe the same syllables, and this converter accepts either.

What hiragana is

Hiragana is one of Japan's three writing systems, and the one everything else is built on. It is a syllabary rather than an alphabet: each of its 46 basic characters is a whole syllable, so か is ka — not a k that needs a vowel attached to it.

Modifying marks extend that set. Two small strokes, the dakuten, voice a consonant: か ka becomes が ga. A small circle, the handakuten, turns は ha into ぱ pa. Writing よ small after a character contracts the two into one syllable, so き ki plus small ょ gives きょ kyo.

Unlike romaji, hiragana is unambiguous. Every character has one reading, which is why it is the first thing learners are taught and the script used for furigana above unfamiliar kanji.

How the converter works

It reads your text left to right and takes the longest syllable it recognises at each step, so sha is read as しゃ rather than as さ followed by は. Four rules sit on top of that:

Small tsu (っ). A doubled consonant marks a brief pause before the next sound, written with a small つ. So gakkou gives がっこう and kitte gives きって. Doubled n is deliberately excluded — nn is the syllabic ん, so konnichiwa gives こんに… and never こっに….

Syllabic n (ん). An n becomes ん unless a vowel or y follows it and binds to it instead. nihon ends in ん; na is な. Where the romaji is genuinely ambiguous an apostrophe forces the split, so kin'en gives きんえん (no smoking) rather than きねん (anniversary).

Long vowels. These are written out as vowels, because that is how the word is spelled in kana: ohayou gives おはよう, not おはよー. The long vowel mark ー belongs to katakana. Macrons and circumflexes expand, so Tōkyō gives とうきょう.

Anything unrecognised passes through. Spaces, digits, punctuation and kana already in your text survive unchanged. Nothing is silently dropped, and converting text twice gives the same result as converting it once.

Examples

RomajiHiraganaThe rule it shows
sushiすしBasic syllables
nihonにほんSyllabic ん at the end of a word
ohayouおはようLong vowel written out
gakkouがっこうDoubled consonant becomes っ
kyouきょうContracted sound
shinkansenしんかんせんん before a consonant
kin'enきんえんApostrophe forcing ん
arigatouありがとうDakuten and a long vowel

One place this stays deliberately literal

konnichiwa converts to こんにちわ, while the greeting is normally written こんにちは.

Both are pronounced the same. The written form uses は because that character is the topic particle, which is spelled ha but said wa. Telling the two apart means knowing that a given wa is a particle, and that is a question about the grammar of the sentence rather than about the sounds in the word.

A converter could special-case the common greetings, but the same sound appears in かわ (river), わたし (I), and any number of names, where わ is correct. Being consistently literal is more useful than being right most of the time and quietly wrong the rest — so type ha wherever you want は, and he wherever you want へ.

When you would use this

  • Learning to write. You can say a word and want to see how it is actually spelled in kana.
  • No Japanese keyboard. You need kana on a machine where you cannot install an input method, for a form or a search box.
  • Checking someone else's romaji. You have a name or a place written in the Latin alphabet and want the kana behind it.
  • Preparing study material. Converting a vocabulary list into kana to practise reading it.

If you are learning to read rather than to type, it is worth working through the kana chart itself at some point — a converter is quicker, but it does the part of the work that is worth learning.

Frequently asked questions

How do I convert romaji to hiragana?

Type or paste the romaji into the box above. The hiragana appears as you type, with no button to press and nothing to install. Both Hepburn spellings (shi, chi, tsu, fu) and Kunrei spellings (si, ti, tu, hu) are accepted and produce the same kana.

Is this converter free?

Yes. There is no sign-up, no usage limit, and no advertising. The site carries no affiliate links or paid placement.

Does my text get uploaded anywhere?

No. The conversion runs in JavaScript on your own device. Nothing you type is sent to a server, logged, or stored — you can disconnect from the network after the page loads and it keeps working.

Why does konnichiwa become こんにちわ instead of こんにちは?

Because こんにちは is written with the topic particle は, which is pronounced wa but spelled ha. Recognising that a particular wa is that particle needs the grammar of the sentence, not just the sounds, so this converter stays literal and always writes わ. Type ha wherever you want は.

How does it handle double consonants like gakkou?

A doubled consonant becomes the small tsu っ, so gakkou gives がっこう and kitte gives きって. Doubled n is the exception — nn is the syllabic ん, so konnichiwa gives こんに and never こっに.

What happens to long vowels such as ohayou?

They are written out as the vowels themselves, so ohayou gives おはよう rather than おはよー. The long vowel mark ー belongs to katakana. If you type a macron or circumflex, ohayō and ohayô both expand to おはよう.

Can it convert hiragana back into romaji?

Not yet. This tool converts in one direction. Anything it does not recognise, including hiragana already in the input, passes through untouched, so running text through twice never corrupts it.

Keep exploring