Consistency Guide
A steady stream is about the spacing between taps, as well as the number of taps. Tap Lab separates pace, alternation and interval variation so that one number does not hide the others.
Read the interval trace
Every point represents an accepted gap between taps. The trace shows the most recent sixty intervals. In Speed test, its center line is the mean of those displayed gaps. In Target tempo, the center line is the chosen target gap. Points above or below the line show longer or shorter gaps. A nearly flat trace suggests even spacing; vertical spikes suggest a hesitation or a fast pair.
Interval variation
The tool computes the population standard deviation of all accepted intervals, then divides it by their mean and expresses the result as a percentage. This is often called the coefficient of variation. Smaller percentages mean more uniform gaps within that run. Identical gaps produce zero percent; that does not prove a perfect physical or musical performance.
The completed result also gives standard deviation in milliseconds. Compare runs made with the same browser, duration and input method. With very few taps, the statistic describes too little data to be useful.
Alternation
Each pair of neighboring taps is a transition. If both taps use the same key, that transition is an error. Alternation percentage is the number of alternating transitions divided by all transitions. The first tap, and the first tap after a pause, has no previous key to compare. A held key does not count as repeated taps.
Early and late estimates
In Target tempo, the first tap anchors a grid with the selected interval. Each later tap is compared with its nearest grid tick. A negative offset is early and a positive offset is late. Mean early/late is the signed average; early and late taps can cancel each other, so read interval variation alongside it.
A tap may be close to a later grid tick after a missed tap. The result therefore also gives target tick coverage: the number of distinct nearest ticks tapped divided by the expected ticks during the active target segments, capped at 100%. Multiple taps nearest the same tick only cover that tick once. Coverage is not an in-game accuracy score.
Browser and sound limits
The browser measures event-arrival time, not the instant a finger physically actuates a switch. Input processing, reduced timer precision, audio output and busy browser scheduling can alter the feedback. The metronome uses scheduled browser-generated sound, but this is not a calibrated end-to-end latency test.
The tool uses a monotonic browser clock, ignores automatic key repeats and pauses when the page is hidden or loses focus. A long rendering interruption also pauses the run. Read the platform details for Performance.now(), keyboard repeats and page visibility.