Chinese to Pinyin Converter

Paste Chinese text to get pinyin with tone marks above every character. Readings are taken word by word, so 银行 and 步行 get the right one — and anything uncertain is flagged instead of hidden.

How it works, and what it cannot do

Pinyin is not a character-by-character lookup. 行 is xíng in 步行 and háng in 银行, and only the surrounding word decides which. This tool splits your text into words first and uses each word's own pinyin from the official HSK 3.0 list; only text outside that list falls back to the most common reading for each character. It does not apply tone sandhi (the tone changes in 不, 一 and consecutive third tones) — you get dictionary readings. It also does not handle traditional characters.

Free to use, no account needed.

Where the readings come from

There is no third-party pronunciation dictionary behind this page. The source is the official HSK 3.0 word list itself: 11,000 entries that each carry the correct pinyin for the whole word. Splitting each entry’s pinyin into syllables and lining them up with its characters yields a reading for 3,021 of the 3,088 characters the list uses — and, more usefully, it yields readings that are already attached to a word rather than floating free.

108 of those characters turn out to have more than one genuine reading (not counting neutral-tone and tone-sandhi variants of the same syllable). Those are exactly the characters a character-by-character converter gets wrong, and they are the reason this tool segments your text into words first.

What it deliberately does not do

No tone sandhi. 不 and 一 change tone before certain syllables, and two third tones in a row shift the first one. You get dictionary readings, which is what a learner should memorise — but it is not what a native speaker says out loud.
No traditional characters. The word list is published in simplified characters only.
No guessing. A character with no reading in the list shows a question mark rather than a plausible-looking invention.

The characters a lookup table gets wrong

These all carry two readings inside the HSK 3.0 list itself, and only the surrounding word decides which one is right. A converter that stores one reading per character has to pick — and is wrong roughly half the time on exactly these, which are among the first few hundred characters anyone meets.

Character	One reading	The other
行	xíng — 行走 to walk	háng — 银行 bank
乐	yuè — 音乐 music	lè — 快乐 happy
觉	jué — 觉得 to feel	jiào — 睡觉 to sleep
长	cháng — 长期 long-term	zhǎng — 校长 headteacher
重	zhòng — 重要 important	chóng — 重复 to repeat
还	hái — 还是 still	huán — 归还 to give back
得	dé — 得到 to obtain	děi — 得走了 must go
差	chà — 差不多 almost	chāi — 出差 business trip
会	huì — 开会 to meet	kuài — 会计 accountant

Where the tone mark goes

Every syllable gets exactly one mark, and its position follows one short rule rather than sitting on the stressed vowel:

If there is an a, it takes the mark — hǎo, xiǎng, kuài.
Otherwise, if there is an o, it does — gōu, duō.
Otherwise, if there is an e, it does — xiè, gěi.
Otherwise it goes on the last vowel — shuǐ, liù, jiǔ.

A neutral-tone syllable carries no mark at all — 爸爸 is bàba, not bàbà. That is a real distinction, not a typographic shortcut: mark the second syllable and you have written a different pronunciation.

A ten-second test for any converter

Paste this into whichever tool you are considering:

他在银行工作，下班后听音乐。睡觉前，他觉得今天很快乐。

Nothing about it is unusual — an ordinary sentence built from HSK 1–3 vocabulary. But 乐 appears twice and needs a different reading each time, and so does 觉. A tool that stores one reading per character is not merely likely to slip here; it is structurally incapable of getting both right. The correct output is:

tā zài yín háng gōng zuò, xià bān hòu tīng yīn yuè. shuì jiào qián, tā jué de jīn tiān hěn kuài lè.

Also free:

Loading…