A one-minute video holds about 140 spoken words when someone reads calmly to camera. That’s the pace RecReview uses for its length estimates, and a reasonable place to start.
Below is a table from 15 seconds to 10 minutes, followed by what moves the number: who’s speaking, in which language, what counts as a word, and the seconds when nobody says anything. For the layout of the script itself, see the two-column script template.
In short
- At a calm read to camera, a minute holds about 140 words: 30 seconds is about 70 and 2 minutes about 280.
- Narrators run faster, interviews run long, and Portuguese needs more syllables than English for the same idea.
- Count only what will be said, the way it will be said, and budget the seconds without speech first.
- Time your presenter reading about 80 words of the real script aloud, and plan with that pace.
- When the script runs long, cut whole sentences instead of asking for a faster read.
Words for each length, from 15 seconds to 10 minutes
The middle column is the 140 baseline. The other two show what the same length holds if your presenter turns out slower or quicker once you time them. Pick a column after the stopwatch test further down.
| Video length | At 120 words/min | At 140 words/min | At 160 words/min |
|---|---|---|---|
| 15 seconds | 30 | 35 | 40 |
| 30 seconds | 60 | 70 | 80 |
| 45 seconds | 90 | 105 | 120 |
| 1 minute | 120 | 140 | 160 |
| 90 seconds | 180 | 210 | 240 |
| 2 minutes | 240 | 280 | 320 |
| 3 minutes | 360 | 420 | 480 |
| 5 minutes | 600 | 700 | 800 |
| 10 minutes | 1,200 | 1,400 | 1,600 |
These are spoken words across the whole running time. A music open or an end card leaves less time for speech, and the count has to shrink with it.
Who’s speaking changes the pace
A prompter keeps people at a steady rhythm, especially when an operator follows them. For a calm read to camera, 140 is a sensible starting point.
It’s also the easiest case to time in advance, since the presenter reads on set exactly what they read at the table.
Professional narrators hold a steady pace for a long time. ACX, Audible’s audiobook production platform, says most narrators read about 9,300 words an hour, roughly 155 a minute.
In a video voiceover the editor can also close the gaps between lines afterwards, which you can’t do inside an on-camera take without a cut.
An interview has no script to count. People answer at conversational pace, with false starts, and the length gets decided in the edit. Counting the answer you expect tells you roughly how much to record, but real answers tend to run long.
English and Portuguese don’t fit the same minute
Words per minute hides the syllables, and syllables are what take time to say. “The new plant opens next week” is six words and seven syllables. The same sentence in Brazilian Portuguese, “A nova fábrica abre na semana que vem”, is eight words and fourteen syllables.
One sentence proves nothing about a whole language, but Portuguese does tend to need more syllables than English for the same idea. A script translated from English usually gets longer when read aloud, even if the word count barely changes.
So don’t carry the English timing over to the Portuguese version, or the other way around. Time each language separately, ideally with the person who will record it.
If a bilingual piece has to land on the same length in both, plan on trimming the Portuguese script, and show the client that trim before the shoot.
Count what will be said, the way it will be said
Only words someone says out loud count. Speaker names, lower thirds, card text, camera directions and music cues stay out of the total.
A written word and a spoken word are different units, which is easy to forget. “42,000 m²” looks like two words and is said as “forty-two thousand square meters”. “$1.5M” becomes “one point five million dollars”.
A web address like fennick.com/careers is one word on the page and five out loud. Acronyms get read letter by letter, so “ESG” takes about as long as three short words.
Write the spoken version in the script and the count gets honest on its own. It also saves the presenter from converting “1.5M” in their head mid-take.
Interview questions are the other trap. Someone reads them on set, but if the editor cuts them, they don’t belong in the length.
The seconds a word count can’t see
A good part of any finished video is nobody talking. There’s the music open before the first line, the product shot that holds for three seconds, the beat after a joke, the chart that needs a moment to be read and the end card with the logo.
None of that shows up in the word count, and all of it comes out of the same running time.
Budget those seconds first and give the speech what’s left. Take a 60-second spot with a 4-second music open, two product shots of 3 seconds each and a 5-second end card.
That’s 15 seconds without speech, so the voice gets 45 seconds, about 105 words at 140 a minute. Write 140 words for it and the edit will have to cut lines or rush the read.
Explainers and tutorials stop for the picture even more often. The narrator waits while a chart builds or the cursor finds the next button, and six graphics that each need about two seconds on their own add 12 seconds the script never mentions.
Time it with a stopwatch, then cut
The number to plan with is your presenter’s, and finding it takes about five minutes.
- Pick a paragraph of the real script, around 80 words.
- Have the person who’ll record it read it aloud at the pace the video needs. If they aren’t available, read it yourself and be honest about how relaxed they’ll be on the day.
- Read it twice and time the second pass, because the first one usually stumbles.
- Then do the division. An 84-word paragraph read in 40 seconds is 84 ÷ 40 × 60 = 126 words a minute, so for that presenter a 90-second video holds about 189 words, not 210.
Reading silently doesn’t work for this. Most people read faster in their head, and the eye skips over the tongue twisters.
When the script comes in long, take out whole sentences. Trimming a word here and there rarely saves more than a second or two, and asking the presenter to speak faster costs you the delivery.
Start with the opening line that warms up before saying anything, the third example when two already make the point, and lines that repeat what’s on screen. If the graphic says “42,000 m²”, the voiceover may not need to say it too.
The length of the video also isn’t the time it takes to record. A 90-second script is about four and a half minutes of recording at three times the final length, and lighting and makeup come on top of that.
The post on building a shoot schedule has the multipliers and a full day worked out.
How this works in RecReview
In RecReview the script lives on the video’s task, as an uploaded file or a document linked from Google Drive or OneDrive.
Calculate duration, in the script’s menu, reads it and strips what isn’t spoken: lower thirds and card text, camera directions, scene headings, sound and music cues and the speaker names. It counts the rest at 140 words a minute.
The task’s Production section then shows the estimated length with the word count next to it. If one document holds the scripts for several videos, the cleanup looks for the part that matches the task’s title.
The count goes by what’s written, so “42,000” counts as one word, and spelling numbers out brings the estimate closer to the real read.
The cleanup keeps interview questions, since someone reads them on set. If they’ll be cut, or your presenter timed at 126 instead of 140, click the length on the task and type the minutes yourself.
When the video goes into a production plan, its block takes that length, including the one you typed if you changed it, and multiplies it by the pace you set for the day: Optimistic 2×, Normal 2.5×, Realistic 3× (the default) or Conservative 4×.
On the teleprompter, editing a block’s text shows the word count and estimated seconds as you type. The presenter’s screen then shows elapsed and remaining time for the block against that estimate as the text scrolls.
Once the cut is uploaded, the task shows the real length of the latest version with the estimate underneath. Over a few videos, the gap between the two tells you how far 140 is from your presenter and from the way your team edits.
