Speaking Time: Words to Minutes and Back

A speech is timed by the words in it and the pace they are spoken at. This guide explains how the reading time calculator turns a word count into a speaking time, how to run the same arithmetic backwards to get a word budget from a time slot, how the speaking constant compares with the silent-reading constant, and why a script full of figures takes longer to deliver than its word count suggests. Every number below is computed by the site's counting engine when the page is built, so the worked examples cannot drift from what the calculator shows.

The formula, in both directions

Speaking time is the word count divided by 150 words per minute. The engine stores the result as a fraction of a minute — 150 words is exactly one minute, half that many is half a minute — and a separate formatting step turns the fraction into the minutes-and-seconds figure on the page. Nothing else feeds the calculation: not the sentence count, not characters, not punctuation. If two texts have the same number of words, they get the same speaking time.

The 150 figure is a common prepared-delivery pace — the rate of someone reading a written script aloud to an audience, neither racing nor pausing for effect. It is an estimate of the middle of a wide range, not a measurement of you, and the calculator offers no alternative rates by age, audience, or occasion because none would be better grounded. The constant sits in the shared counting engine next to the silent-reading rate, which the reading time guide covers in its own right.

Because the formula is a single division, it reverses cleanly. A time slot multiplied by 150 gives the number of words that fill it at the assumed pace: minutes × 150 = words. Speakers need that reverse direction most often, since a call for talks gives minutes rather than a word limit, and the drafting question is how much to write.

Worked example: a toast, words to seconds

The engine finds 49 word-tokens in that toast. Dividing by 150 gives 0.327 of a minute; multiplied by 60 that is 19.6 seconds, which the formatter rounds to the nearest whole second and shows as "20 sec". The same 49 words divided by the silent-reading rate of 238 come out at "12 sec" — the reader is faster than the speaker. Note what did not matter: the toast has commas, 3 sentences, and proper names, and none of them changed the result, because only the word count enters the division.

Worked example: from a time slot to a word budget

Turn the formula around and a slot becomes a target. The table lists a handful of slot lengths and the words that fill each one at 150 spoken words per minute, with the silent-reading budget at 238 beside it for contrast. Each row is one multiplication, rounded to the nearest word.

Read the 45 min row as a drafting target: a script of about 6,750 words fills the slot with no time for anything but talking. Whatever you plan to spend on a pause, a demonstration, a question from the floor, or the walk to the lectern comes out of that word budget, not in addition to it — so most speakers should draft under the figure and let rehearsal decide how far under. The calculator makes the trimming visible: paste the draft, watch the speaking-time figure, and cut until it sits at or below the slot.

Reading and speaking, side by side

The two constants describe different activities. Silent reading covers 238 words in the time speaking covers 150, so the reading figure is always the shorter of the two by the same fixed ratio, 1.59 to one — or, put the other way, a listener spends about 59% longer with a text than a silent reader does. This is why a 1,000-word piece shows "4 min 12 sec" in one box of the calculator and "6 min 40 sec" in the other. Choose the box that matches how the text will be consumed: an article, a newsletter, or a report is read; a talk, a toast, a voice-over script, or a recorded lesson is spoken. Both figures carry the same caveat — each is the centre of a wide range — and the methodology page states the sources so you can weigh them yourself.

Why number-heavy scripts run long

The engine's word rule is a run of letters, digits, apostrophes, and hyphens; anything else — including a comma or a period — ends the token. That rule is what makes "state-of-the-art" and "it's" one word each, and it has a consequence for figures. The string "2,500" contains 2 tokens, because the comma splits the digits into separate runs. Said aloud, it becomes "two thousand five hundred", which is 4 words. A decimal behaves the same way: "3.14" is 2 tokens on the counter and 4 words in the mouth ("three point one four"). This is a property of the tokenizer, not an error in it — the counter is counting what is written, and a written figure is shorter than its spoken form.

The two passages are the same sentence, once with figures and once as a speaker would voice it. The figures version counts 22 words and estimates 9 sec; the voiced version counts 28 words and estimates 11 sec. The difference is small in one sentence and compounds across a script — a quarterly update, a results announcement, or a lecture with dates and measurements in every paragraph will run past its estimate for exactly this reason. Two practical responses: draft such a script with the numbers written out as words, so the counter sees what the audience will hear, or keep the figures and treat the speaking estimate as a floor. Even the voiced version above is a slight under-count, since the hyphen in "twenty-five" keeps it a single token while a speaker voices both halves — the same joining rule that keeps "state-of-the-art" whole.

The opposite distortion exists too, and the what counts as a word guide covers it in full: a pasted web address such as https://example.com/guide/counting contributes 5 word-tokens because its colons, slashes, and dots all separate runs, and a bare hyphen used as a bullet registers as a token on its own. Speaker notes that carry links, slide file names, or bulleted fragments will therefore estimate slightly long. For an ordinary prose script neither effect is large; for a script that is mostly figures or mostly links, it helps to know which way the estimate leans.

How the seconds are displayed

The engine keeps the speaking time as an exact fraction of a minute and rounds only at display time. The formatter multiplies the fraction by 60, rounds to the nearest whole second, and then splits that into minutes and seconds: a value that lands on an exact minute shows no seconds part, a value under a minute shows seconds alone, and anything else shows both. Very short inputs get a phrase rather than a figure — "Thank you all." is 3 words, and its speaking time displays as "under 5 sec", because a figure of a second or two would promise a precision the estimate does not have. Empty text shows "0 sec". Rounding happens once, on the final figure, so no error accumulates across a long text: a script is timed as a whole, not sentence by sentence.

What the estimate cannot see

The formula is honest about what it models — continuous delivery of the words on the page at a steady pace — and silent about everything else. It cannot see the pause after a question, the seconds spent advancing a slide, laughter, applause, a demonstration, or the walk from the chair to the microphone. It cannot see whether you speak quickly or slowly, and there is no setting for pace. What it does reliably is comparison: two drafts of the same talk can be compared to the second, and a script that estimates well over its slot before rehearsal is certainly over it. Draft to the budget, then rehearse with a clock: the calculator tells you how much you wrote, and the clock tells you how long you take to say it. The other guides cover the counting rules the word figure itself depends on.

Frequently asked questions

Counting happens locally in your browser — nothing you type is transmitted or stored. Every figure in this guide was computed at build time by the same engine the calculator runs, and both pace constants are estimates of the middle of a wide range. See the methodology page.