Typing speed test & word-attack typing game
Classic mode measures WPM or characters per minute; switch to Word Attack (单词来袭) — words fall from the sky, type them to pop them and build combos, and the speed test turns into a game.
开始打字即自动计时;写错的先高亮红、改对即转绿。中文模式用输入法正常打,上屏才计入。
单词从上方掉落。敲击某个单词的首字母锁定它,继续敲完整词即击爆; 打对不断连击、每 8 个词升一关(更快、更多);放一个词落地掉一颗心,三颗心掉光结束。 中文拼音模式:看到汉字、敲它的拼音。三档难度 = 词库取前 100 / 500 / 1000 词,同时决定下落速度与同屏数量。
结算口径与经典模式一致(英文 WPM ÷5 字符计词、中文按每分钟正确汉字),所以这本质上还是一台测速仪——只是好玩。手机可玩,实体键盘成绩更有可比性。
WPM 是什么?打字多快算"快"?
WPM 的国际口径是"每 5 个正确字符算 1 词"。英文打字测试通行的 WPM(Words Per Minute)=(正确字符数 ÷ 5)÷ 分钟数——5 个字符(含空格)折算一个"标准词",这样短文长文都可比。本页同时给准确率和错误数,速度高但准确率掉到 90% 以下的,实际岗位效率并不会高。
中文不看 WPM,看"每分钟汉字数"。中文经由输入法上屏,一个汉字的击键成本远高于一个英文字母,所以中文测速通行口径是每分钟正确上屏汉字数(对应老式速录考核的"字/分钟")。一般办公场景 60~80 字/分已经够用,专业录入 120+ 字/分。
大致参照(60 秒测试)。英文:30 WPM 生手,40~50 普通办公,60~70 熟练,80+ 明显高于平均;键盘盲打是跨过 40 之后的必修课。简历上想写"打字快",先测一次拿到自己的 WPM——比形容词可信。
准确率怎么算。把已输入与目标文本逐位对齐:位置相同算正确,不同算错误,多打又删回来的会被改对(转绿);未删除的多余尾串计为错误。错误数计入分母之外单独展示,方便你回看是"快而糙"还是"稳而慢"。
练不下去,就换个玩法。打字是典型的"平台期很长"的肌肉记忆训练,枯燥的跟打练习大多坚持不过两周。本页的「单词来袭」模式把同样的击键藏进游戏里:单词掉落逼你看词整体而不是逐个字母、连击倍率逼你少回头删改、关卡提速逼你在临界速度上保持准确率——这三件事恰好是从 50 WPM 迈到 70 WPM 的关键。玩一局,结算面板照旧给你 WPM 和准确率,练没练到看得见。
关于本页的数据。目标文本与游戏词库都内置在页面里,你输入的内容只在浏览器内存里比对——不发起任何网络请求,不写 localStorage、cookie 或网址参数,刷新即清空;音效也是浏览器现场合成的,全站零外部资源。
FAQ
No. WPM converts by '5 characters = 1 word'; KPM (keystrokes per minute) counts raw keystrokes, usually higher than WPM and including the wrong ones; Chinese tests score correct Chinese characters per minute. The three conventions cannot be converted into each other - state the convention before quoting a score. This page gives WPM for English and characters-per-minute for Chinese.
Just fix it: delete and retype - corrected characters flip from red to green and stop counting as errors. Scoring counts only characters still wrong at the moment of settlement, so fixing as you go matches real office typing and costs you no penalty.
No. Chinese mode brings characters into the comparison only after the input method commits them (compositionend); the pinyin sitting in the candidate box does not participate. Type normally - choosing candidates will not interrupt anything.
You can, but scores on a phone's virtual keyboard are not comparable with physical keyboards - plug one in if you can. Results are for self-reference only: sites differ slightly (word-length definition, error policy), so cross-site score comparison means little.
Words fall one by one from the top; typing a word's first letter locks onto it (it gets an orange frame), then finish the word to pop it for points. Every 8 words raise the level: faster fall, more words on screen; one word landing costs a heart, three lost hearts end the run. Popping in a row builds a combo multiplier (up to x10); one wrong keystroke breaks it. At start you pick difficulty: Easy / Medium / Hard take the first 100 / 500 / 1000 words of the list, longer and rarer, with fall speed and on-screen count rising too. The summary panel reports speed on the same conventions as classic mode (English WPM; the Chinese pinyin mode by correct letters per minute) - essentially a speed test under cover of pressure: the falling words force you to type while thinking, closer to real office rhythm.
Two word pools: English words, and Chinese pinyin - you see the character and type its pinyin as shown under it (for the character meaning 'overtime at work', type jiaban). Every pinyin is laid out under its character for you to copy, and you pick up the correct spelling along the way; if that is not your thing, just play the English pool.
Nothing is downloaded. The sounds (pop / miss / level-up) are synthesized on the spot by the browser's WebAudio, with one-click mute; words are plain text with a CSS glow - no images, no external resources at all. The whole page stays upload-free: game data lives only in this page's memory, refresh clears it, no scores are stored and there is no leaderboard.
No. The page makes zero network requests (the analytics script records only that the page was visited, never what you typed); all comparison happens in browser memory, nothing is written to local storage, and a refresh clears everything.