The Science of 120–160 Words per Minute
Decades of measurement converge on a narrow band: comfortable English speech for a listening audience runs at roughly 120 to 160 words per minute. Everyday conversation is faster, often 160 to 200 wpm, because dialogue lets a confused listener interrupt. Presentation speech must be slower — the audience gets exactly one pass at every sentence, and comprehension research consistently shows retention dropping as rates push past the mid-100s for unfamiliar material. Professional norms cluster inside the band: audiobook narrators are directed to 150–160 wpm, US radio announcers read news around 150–175, and TED coaches nudge speakers toward the slower half for idea-dense talks.
The great speeches sit strikingly low. Analyses of Martin Luther King's "I Have a Dream" put its opening sections near 90–100 wpm, accelerating as the speech builds. Rapid speakers exist — John F. Kennedy was once clocked near the top of the range — but note the direction of the exceptions: speakers slow down for weight, not speed up.
Character-counted languages have their own equivalents. Korean broadcast and presentation guidance falls around 250–300 syllables (roughly characters) per minute; Japanese NHK announcer training famously targets about 300 characters per minute for news reading; Mandarin news reading commonly runs 240–300 characters per minute. That is why this tool's normal setting sits at 130 wpm or 270 cpm: it is the planning midpoint of what real, comprehensible delivery looks like. Nervous speakers reliably drift 10–20% faster than rehearsal, so a script timed at the "normal" rate leaves exactly the cushion you will need on stage.
Reference speaking rates (English, words/min):
audiobook narration 150-160
US radio news 150-175
conversational speech 160-200
presentations (target) 120-160
MLK "I Have a Dream" ~90-100 (opening)
Character-counted languages (chars or syllables/min):
Korean broadcast ~250-300
Japanese NHK news ~300
Mandarin news ~240-300
This tool: slow 100 wpm / 200 cpm
normal 130 / 270 <- planning midpoint
fast 160 / 340From Word Count to Wall Clock: Budgeting a Talk That Fits Its Slot
The raw conversion is simple division — 1,950 words at 130 wpm is 15 minutes — but a talk is not a read-through, and the difference is where schedules die. Everything you do that is not speaking still consumes the slot: advancing and orienting on slides (comfortably 5–15 seconds each), a live demo, drinking water, waiting out laughter, fielding a mid-talk question. Slide-heavy decks are the biggest hidden cost; that is why this tool lets you add per-slide minutes on top of the text estimate rather than pretending narration is the whole story.
Working speakers budget backwards from the slot instead of forward from the script. A robust rule is to plan content for about 90% of the allotted time: for a 20-minute slot, script roughly 18 minutes, or about 2,300 words at 130 wpm. If questions are taken inside your slot — standard at academic conferences, where a "15-minute talk" is often 12 minutes plus 3 for Q&A — subtract that first, before the 90% rule. Chronic overrun has a social cost that is easy to underestimate: at any multi-speaker event, your extra four minutes are stolen from the next presenter or from the audience's break, and session chairs remember.
Build in a cut plan as well. Mark one section of the script as droppable in advance — a second example, a digression, an extra case study — so that when the 5-minutes-left card appears you excise cleanly instead of accelerating into a mumble. Speeding up rescues almost nothing: a 15% rate increase saves barely two minutes on a 15-minute talk while measurably damaging comprehension. Cutting content works; compressing delivery does not.
Budgeting a 20-minute conference slot:
slot 20:00
Q&A inside slot? - 3:00
safety margin (90%) - 1:42
speakable time ~15:18
words at 130 wpm = 15.3 x 130 = ~1,990 words
15 slides x 15s of transitions = 3:45 <- comes OUT of the 15:18
net narration ~11:30 -> script ~1,500 words
Speeding up does not save you:
1,990 words at 150 wpm = 13:16 (saves ~2 min,
costs comprehension) -> cut content insteadRehearsal That Actually Works: Why Silent Reading Lies to You
The most common timing mistake is rehearsing with your eyes. Silent reading runs at 200–300 words per minute — roughly double speaking pace — so a script that "reads in eight minutes" is a fifteen-minute talk. Worse, silent reading skips everything that makes delivery slow: breathing, pausing, walking, clicking, and the small recoveries after a stumble. The only rehearsal that produces a trustworthy number is out loud, standing, at performance volume, with the actual slides. That is what the countdown timer on this page is for: paste the script for the estimate, then verify it with a spoken run.
Rehearsal quality follows well-studied learning principles. Spaced practice beats cramming — three run-throughs across three days outperform six in one evening, a spacing effect replicated across a century of memory research. The testing effect applies too: reciting from memory and checking afterward strengthens recall far more than re-reading. Record one full run on your phone; speakers systematically mis-hear their own pace and filler-word rate ("um", "so", "like" — or 그, 어, えーと, 那个), and a recording is the cheapest honest coach available.
How many full rehearsals is enough? Coaches converge on a range: below three, timing estimates are unstable; somewhere past ten, delivery risks flattening into recitation for non-professionals. A practical ladder: first run for content (expect it to be ugly and long), second and third against the clock, cutting after each, a fourth in front of one honest listener. Rehearse the opening and the closing disproportionately — they anchor the audience's impression, and they are what nerves attack first.
Why silent reading misleads:
2,000-word script
silent reading ~250 wpm -> 8 min "feels fine"
spoken delivery ~130 wpm -> 15:23 actual talk
+ slides, pauses, nerves
Rehearsal ladder (spaced over several days):
run 1 content pass (will run long - fine)
run 2 vs the clock cut what overflows
run 3 vs the clock cut again, record it
run 4 live listener fix what they flag
extra opening + closing until automaticThe Pause Is Part of the Speech: Silence, Fillers, and Breathing
Word-per-minute numbers hide a surprising fact: in natural speech, silence occupies a large share of the clock. Pause research distinguishes short syntactic pauses at clause boundaries (a few tenths of a second), sentence pauses (around half a second to a second), and rhetorical pauses — the deliberate two-to-three-second silences great speakers deploy before or after a key line. Audiences consistently rate speakers who pause as more confident and more intelligible, yet from the stage a two-second silence feels like an eternity; closing that perception gap is one of the main things rehearsal buys you.
Pauses also do mechanical work. They are when you breathe — running out of air mid-sentence forces audible gasps and rising pitch — and they are when the audience thinks. Cognitive load research on lectures suggests listeners need processing gaps after dense propositions; a pause after your key claim is not dead air, it is the time in which the claim lands. Speakers who fear silence backfill it with fillers, and the fix is mechanical rather than moral: fillers spike at planning boundaries, so knowing your transitions cold removes most of them. Recording yourself once (as suggested above) typically reveals a filler rate two to three times higher than you would have guessed.
Rate variation is the final pacing instrument. A uniform 130 wpm is comprehensible but hypnotic; skilled speakers slow to near 100 for definitions and conclusions, and let anecdotes run quicker. The practical encoding: mark your script — underline what must be slow, bracket what can be fast — and let the average, not every sentence, hit the target rate this tool assumes.
Formats Around the World: TED, PechaKucha, Lightning Talks, and the Academic 15
Time limits are not arbitrary — each famous format encodes a theory of attention. TED caps talks at 18 minutes, a length its curators defend as long enough for a serious idea and short enough to hold undivided attention. PechaKucha, invented by architects in Tokyo in 2003, is stricter still: exactly 20 slides that auto-advance every 20 seconds, 6 minutes 40 seconds total, a format designed to cure architects of rambling. Its corporate cousin Ignite uses 20 slides at 15 seconds. Lightning talks at tech conferences run 5 minutes with a hard cut; academic conferences typically allot 12–20 minutes including questions; a Japanese business 報告 or a Korean 업무보고 may be expected to land in three. Toastmasters, the global speaking club, builds its whole pedagogy around 5-to-7-minute speeches with visible traffic-light timing.
The uniting lesson is that constraint improves talks. Auto-advancing formats force rehearsal because winging it is impossible; short formats force ruthless prioritization of one idea. Speakers who train on a 6:40 PechaKucha routinely report their 30-minute talks improving too.
To practice any of these with this page: type the format's total time as a manual count trick — or simply paste your script, pick a speed, and check the estimate against the format's ceiling before you rehearse against the countdown. If the estimate says 8:30 for a lightning talk, do not rehearse hoping to go faster; cut a third of the script first. The timer will tell you the truth either way; the only question is whether you hear it before the audience does.
Famous formats and their word budgets (at ~130 wpm):
PechaKucha 20 slides x 20s = 6:40 -> ~850 words
Ignite 20 slides x 15s = 5:00 -> ~650 words
Lightning 5:00 hard stop -> ~650 words
Toastmasters 5-7 min -> 650-900 words
Academic 15:00 (12 talk + 3 Q&A) -> ~1,550 words
TED 18:00 max -> ~2,300 words
Workflow on this page:
paste script -> read estimate -> over the ceiling?
cut FIRST, then rehearse with the countdown timer.