TYPING GUIDE

How typing speed and accuracy are calculated

Understand gross and net WPM, accuracy, reference-prefix errors and Hindi scoring with checked examples from the typing practice tool.

What the result measures

The typing practice tool measures the answer committed to its input box and the elapsed time. It does not count every physical key press. Backspace, Shift, candidate selection and a corrected mistake therefore do not each become a new typed character in the final score. The result describes the text that remains when you finish. Use it to compare your own practice under similar settings, not as an official examination mark.

Gross WPM: the five-character convention

Here, one standard word means five Unicode code points after NFC normalisation. Spaces, capitals and punctuation are part of the text. Gross WPM is characters typed divided by five, divided by elapsed minutes. The displayed WPM is rounded to a whole number.

For example, 600 characters in 60 seconds gives 600 / 5 / 1 = 120 gross WPM. If the same text takes two minutes, it gives 60 gross WPM. This convention measures the amount of text, so a passage with many short dictionary words is not scored by counting those words individually. Very short attempts can produce unusually high speeds; use a sustained attempt before drawing a conclusion about your normal pace.

Net WPM and accuracy: a worked example

The tool compares your answer with a reference prefix using edit distance. A substitution, insertion or deletion is one edit in the best alignment. Net WPM is the typed character count minus edit errors, divided by five and elapsed minutes, with a minimum of zero. Accuracy is 100 multiplied by one minus edit errors divided by the longer compared string, also limited to a minimum of zero. Accuracy is displayed to one decimal place when needed.

For a 600-character answer in one minute with 12 edit errors, the result is 120 gross WPM, 118 net WPM after rounding, and 98% accuracy. This is an example of this website's formula, not a universal penalty rule. Another site or an examination authority can use a different denominator or error classification and produce a different score for the same answer.

Why the reference prefix matters

The reference is shortened to the length of your answer before comparison. Typing “hello” against “hello world” compares only “hello”. The remaining untyped words do not count as errors. At sixty seconds this example gives five typed characters, zero errors and 1 WPM. A high accuracy percentage by itself does not show that you completed much of the passage.

The prefix convention also has a limitation. With reference “abcdef” and answer “abdef”, the comparison uses “abcde”, because the answer has five characters. The best alignment counts two edits: the missing “c” and the extra final “f”. This can look different from a full-passage comparison. Inspect the final character comparison before interpreting an unexpectedly high error total.

Live highlights and final error review

The highlight while you type checks positions, which keeps feedback responsive. Missing a character can make several later positions appear wrong. The final review uses alignment, so its errors can differ from the running positional highlight. Corrected errors are absent from the final answer, although the time spent correcting them still affects speed. Switch backspace off only when that is the practice condition you want to measure.

An empty answer can show the scoring function's initial 100% accuracy value while reporting zero characters and zero speed. It is not a successful completed attempt. Always read characters, time, speed and accuracy together.

Hindi text and fair comparisons

A visible Hindi syllable can contain multiple code points, and an input method can use several keys to commit one unit of text. NFC normalisation makes some equivalent encodings comparable, but it does not convert legacy fonts or treat different words as equivalent. Read the Hindi input guide for setup and composition checks.

Compare unfamiliar passages using the same language, input method, duration and correction setting. After a result, choose one repeated mistake, practise it slowly and retry with fresh text. A remembered paragraph or predictive keyboard may improve the number without measuring the same skill. You can save results on this device or download your report; neither action creates an official qualification.