How many words per minute do people speak on YouTube?

A common rule of thumb for a spoken script is 150 words a minute. It is a fair figure for a prepared talk. We suspected it was a poor one for much of what people watch on YouTube, so we measured it.

On 22 October 2026 we pulled the caption tracks of 107 English YouTube videos across 11 genres, counted the words in each, and divided by the video's runtime. This page is the result, the method, and the ways the numbers can mislead you.

How we measured

For each video we asked the YouTubeScribe transcript extractor for its caption track. We took every caption line, removed sound labels such as [Music] and the music note symbol, split what was left on spaces, and counted the pieces. We divided that count by the runtime YouTube reports for the video, in minutes.

We ran one search per genre and picked videos from the results: a TED talk search, a full lecture search, a podcast interview search, a Python for beginners search, a BBC News search, a laptop review search, a narrated history documentary search, a sleep history search, a stand-up special search, a 10 minute guided meditation search and a Minecraft commentary search. We kept English tracks only, which dropped three Hindi tech reviews. One stand-up special was age restricted and could not be read, so that genre has 9 videos, not 10. Rerun the same searches and you will get different videos, so treat the numbers as a snapshot.

For comparison we added the 60 Shorts that had an English track with at least one word from the sample in our post on whether Shorts have captions. They went through the same count.

Words per minute, by kind of video

Each row is the median of the videos in that genre. The middle half shows where the central half of the videos fell, and the range shows the slowest and fastest single video.

The TED row is 12 talks from one channel and the news row is 12 reports from one channel, because we asked for those channels by name. Treat those two as the pace of a house style, not of the whole genre.

What the numbers say

Three groups show up. The slow group is deliberate speech: guided meditation (55), narrated documentaries (108) and sleep narration (131). The middle group is prepared or performed talk: TED talks (149), stand-up (154), lectures (160) and the BBC reports (168). The fast group sits between 183 and 189: podcasts, gaming commentary, Shorts, tech reviews and coding tutorials.

The spread inside a genre matters as much as the median. Lectures had a middle half of 119 to 170, and one lecture came in at 78 words a minute. Gaming commentary ranged from 83 to 261. We did not check why, but the runtime includes any stretch where the player is not talking. A single number per genre is a guide, not a promise.

For context, the Wikipedia article on words per minute gives 150 to 160 words a minute for audiobooks and 100 to 125 for a comfortable slide presentation. It cites a 1998 conference paper and a 2014 study-skills book, and we did not open either. Our TED row (149) sits just under the audiobook range, the lecture row (160) at its top edge, and the documentary row (108) inside the slide range. That is some reassurance that the method is not off by a large factor.

How many words fit in 2, 5 and 10 minutes

This is the number most people search for. The table is arithmetic from the medians above, rounded to the nearest word, so it answers the question for four kinds of video.

Run it the other way for a script. 1,000 words lasts about 9 minutes 16 seconds at documentary pace, 6 minutes 43 seconds at TED pace, and 5 minutes 28 seconds at podcast pace. At meditation pace it lasts 18 minutes 11 seconds.

For Shorts the same sum gives about 185 words in a 60 second Short and about 550 in a three minute one at the 184 pace. Real Shorts are far shorter than that. The median Short with a track held about 85 words in roughly 30 seconds, which is the same pace as the table. Our guide to how long a Short can be has the limits.

What can mislead you in these numbers

• The sample is small. 5 to 12 videos a genre, found by one search each on one day. The medians are a snapshot, and sleep narration (5 videos) and documentaries (7) are the thinnest.
• We divided by the whole runtime. That includes pauses, music and intros. As a check we divided instead by the time from the first caption line to the last. Every genre median moved by 2 words a minute or less (TED 149 to 151, podcasts 183 to 184, meditation 55 to 56), so intros and outros are not driving the result.
• We counted caption words, not spoken words. 80 of the 107 tracks were automatic captions, which YouTube says can misrepresent what was said. The 27 uploaded tracks gave a lower pooled median (149) than the automatic ones (167), but all 12 TED talks are in the uploaded group, so that gap may be mostly genre. We did not test accuracy.
• Automatic caption lines overlap in time, so we checked that the text did not repeat. We checked two automatic tracks and found no repeated runs of words, so overlap does not appear to inflate the counts. That is a spot check, not proof for all 80.
• A word is whatever sits between spaces. A number like 2026 or a hyphenated word counts once.
• English only. Other languages have different word lengths, so the figures do not transfer.

Choosing a pace for your own script

If you are timing a script, start from the genre you want to sound like. Calm, prepared delivery to camera is near 150. A relaxed, talk-over-the-screen tutorial or review is near 185 in practice. A documentary read or a bedtime narration is near 110 to 130. Type your word count and a pace into the Words to Minutes Converter to get the runtime.

If you have a full script, paste it into the Script Timer. It shows the runtime at slow (110), normal (150) and fast (190) paces next to the pace you pick, plus word, character and sentence counts. Every genre median we measured sits between 108 and 189 except guided meditation, so the three presets cover nearly all of them.

The most reliable pace is your own. Take a finished video, divide the words in its script by its runtime in minutes, and use that number next time. Pace also varies a lot inside one genre. The middle half of our gaming videos spanned 110 to 197 words a minute, so measure yourself rather than borrow a median.

Questions people ask

Sources

1. YouTubeScribe measurement, 22 October 2026. 107 English videos from 11 genre searches and 60 English Shorts, each read through the transcript extractor. Words counted from caption tracks and divided by runtime.
2. Wikipedia. Words per minute. en.wikipedia.org· Checked 22 October 2026. Secondary source: it cites Williams (1998) and Wong (2014), which we did not open.
3. YouTube Help. Use automatic captioning. support.google.com· Checked 22 October 2026.

Reviewed on 22 October 2026. If a line is wrong, email support@youtubescribe.com.