Measure your real typing speed in 15, 30, or 60 seconds, then find out which keys are actually costing you the time. Every miss is tallied against the character the passage asked for, so the test finishes with a ranked list of the keys to drill rather than a single number to feel vaguely bad about. Eighty passages across four sets mean the text keeps changing.
Step by step
01Pick a clock and a passage set
Fifteen seconds is a sprint, sixty is closer to sustained writing. The passage set decides what you are practising: plain prose, punctuation-heavy sentences, the number row, or code with brackets and operators.
02Click the passage and start typing
The timer starts on your first keystroke, not when the page loads, so there is no penalty for reading ahead. Correct characters turn solid, mistakes are marked in place, and the caret underlines the character you are on.
03Read the weak-key report
Below the passage, keys are ranked by how many times you missed them and what share of their appearances that represents. A key has to be missed repeatedly, or appear often enough to be measured, before it is flagged — one slip is noise.
04Run the same passage again
The retry button keeps the exact text and resets the clock, which is the only honest way to see whether a second attempt was faster or just easier. New passage rolls a fresh seed instead.
Why this one
It names the keys, not just the score
Accuracy as a single percentage tells you that something went wrong and nothing about what. This test counts every expected character and every miss against it, so the report says 'semicolon: 4 misses in 9' — a thing you can practise.
The passage is reproducible
Text is generated from a seed carried in the URL. Sharing the link hands someone the same test rather than the same settings, and the retry button re-runs the identical passage so two results are genuinely comparable.
Four passage sets, not one
Prose practice never trains the number row or the bracket keys, which is why programmers and bookkeepers often type well below their prose speed at work. The punctuation, numbers, and code sets put those characters under the clock — twenty passages in each, so a set does not become familiar.
Backspace is a rule you choose
Leaving corrections on measures how you actually write. Turning them off measures first-pass accuracy, because a mistake stays on screen and in the count. Both are useful; they are just different tests.
WPM computed the standard way
Correctly typed characters are divided by five — the conventional word length, including spaces — and scaled to a minute of elapsed time. Wrong characters earn nothing, so rushing cannot inflate the number.
No account, no leaderboard
Your best speed, best accuracy, and session count live in this browser's localStorage. There is no score API, no sign-in, and nothing to compare yourself against except your own last run.
FAQ
Around 40 words per minute is a common average for people who type daily without formal training. Sustained 60–70 WPM is quick, and above 90 is rare outside transcription work. Speed is only meaningful alongside accuracy: 70 WPM at 88% accuracy usually means more time lost to corrections than 55 WPM at 98%.
Accuracy, and it is not close. Every wrong character costs the keystroke that made it plus the keystroke to remove it plus the break in rhythm. Aim for 96% or better at a comfortable pace, and speed tends to follow on its own as the movements stop needing attention.
Correct characters divided by five, then divided by elapsed minutes. The five-character convention counts spaces, so it is not the same as counting real words — it is the standard that makes results comparable between tests. Only characters that matched the passage count toward it.
Brackets, quotes, underscores, and the number row sit outside the home-row patterns that prose drills. Most people lose 20–40% of their prose speed on code, which is exactly why the set exists: the weak-key report will usually point at the same three or four symbols every time.
Yes. The retry button keeps the passage and resets the clock. A new passage is only rolled when you ask for one, or when you change the duration or the passage set — the text length is tied to the clock, so those settings have to regenerate it.
In this browser, in localStorage, under a single key. They do not sync to your other devices, they are not uploaded anywhere, and clearing site data removes them. Share links carry only the duration, passage set, backspace rule, and seed.
The test runs entirely in your browser. Keystrokes are never transmitted, results stay in localStorage on this device, and the share link carries only the settings and the passage seed.
Keep going
Mental math gym
Timed arithmetic sets with a per-skill accuracy breakdown.
Code reading gym
Predict JavaScript output, then trace what actually ran.
Map memory
Countries, capitals, and flags on an interactive world map.
Word counter
Reading time, readability, and keyword density for any text.