Chinese to pinyin
Characters to pinyin with tones, numeric tones, no tones or initials · polyphonic characters resolved by phrase · custom readings
要转换的中文
夹带的英文、数字、标点、emoji 默认原样保留（可在下面关掉）；单次建议不超过 20 万字符。
输出模式
转换结果
等待输入
逐字对照表只列汉字
| 原文 | 带声调 | 数字声调 | 无声调 | 首字母 | 其它读音 |
|---|---|---|---|---|---|
| 还没有内容 | |||||
对照表最多显示前 2000 个汉字，完整的用「下载 CSV」。
自定义读音（多音字纠正）
每行一条，写成「词=拼音」，拼音用带声调的字母写法（例如 重庆=chóng qìng）。自定义优先于内置词典：定义了单字「重=zhòng」会连「重庆」一起改掉，所以建议直接定义词组。
输出规则与已知边界
| 模式 | 「中文」的输出 | 适合 |
|---|---|---|
| 带声调 | zhōng wén | 注音、教学、给汉字标音 |
| 数字声调 | zhong1 wen2 | 程序处理、排序与检索 |
| 无声调 | zhong wen | 做 URL、文件名、检索键 |
| 首字母 | z w | 缩写、首字母检索 |
多音字：默认按词组上下文取音（重庆 chóng qìng、重量 zhòng liàng、银行 yín háng、长大 zhǎng dà），词组词典在本地；脱离词组只能取常用读音（单独的「重」读 zhòng）。词典并非完美：个别词条会取音不准（实测「行长」会被读成 háng cháng），遇到这种直接用下面的「自定义读音」指定即可。开启「列出全部读音」会把多个读音并列输出。
逐词分隔：用自带的本地小词典切词，词内不加分隔符、词间加分隔符。词典覆盖有限——常用词（重庆 / 银行 / 音乐）会成组，未收录的词（例如「你好」）仍会被逐字切开，所以逐词结果有时与逐字接近；切分只决定分隔位置，读音不受影响，边界对不上时会自动回退成逐字分隔。
空白与换行：原文里的空格、换行、制表符都原样保留；它们同样算「非汉字」，两侧各会有一个分隔符，所以输出里可能出现连续空格。想要更干净的结果，可以关掉「保留非汉字」或把分隔符设为「无」。
ü 与 v：只有无声调、数字声调模式支持 ü → v（女绿 → nv lv）。带声调模式保持 ǚ / ǜ，该模式下不提供这个开关。
变调：默认按普通话习惯处理「一 / 不」的变调（一个 → yí gè、不是 → bú shì）；关掉后输出本调（yī gè、bù shì）。三声连读变调不做（你好 输出 nǐ hǎo）。
生僻字与库边界：读音来自本地词典，生僻字按库内数据（龘 dá、靐 bìng、齉 nàng）；词典覆盖不到的字符会原样保留。emoji、日文假名、其它符号都不是汉字，按「非汉字」处理。
FAQ
By default, from the surrounding word context: 重庆 reads chóng qìng while 重量 reads zhòng liàng, 银行 reads yín háng while 长大 reads zhǎng dà - these are resolved locally by a built-in phrase dictionary, without the network. Out of a phrase only the common reading can be taken (a lone 重 reads zhòng), so writing words in full rather than splitting them makes the judgement clearly more accurate. Turning on 'list all readings' outputs every reading of the character (重 → zhòng chóng) so you can confirm yourself.
One per line, as word=pinyin with tone-marked letters (e.g. 重庆=chóng qìng); numeric tones (zhong4) are not accepted - the dictionary only knows the tone-marked form. Custom entries take priority over built-in phrases: if you define the single character 重=zhòng, then even 重庆 follows it and becomes zhòng qìng, so defining the phrase directly (重庆=chóng qìng) is recommended. Lines that fail to parse are reported line by line and do not take effect; 'restore default readings' clears all custom entries.
Take 中文: tone-marked is zhōng wén (good for annotation, teaching, marking hanzi), numeric tone is zhong1 wen2 (good for programmatic processing, sorting and lookup), toneless is zhong wen, initials are z w (good for abbreviations and initial-letter lookup). Writing ü as v is only in effect in the toneless and numeric modes: 女 绿 → nv lv; the tone-marked mode keeps ǚ / ǜ unconverted and this tool offers no such switch in that mode (the result would be non-standard).
No. The dictionary and conversion all run in your browser: the text never leaves your machine, refreshing clears it, the page makes no network requests, and there is no third-party CDN dependency. Up to 200,000 characters per run is recommended (beyond that you are asked to split); in measurement, 100,000 hanzi take about 150 milliseconds. Conversion is pure local string processing and uses no server resources. English, digits, punctuation and emoji mixed into long text are not hanzi and are kept as-is by default; you can also turn off 'keep non-hanzi' to output pinyin only.