Caption Reading Speed & Line Length: The Rules
Captions have a speed limit. Netflix caps subtitles at roughly 17 characters per second for adults and keeps lines to about 42 characters (Netflix English (USA) Timed Text Style Guide); its general spec never shows more than two lines at once and holds each cue on screen for at least five-sixths of a second (Netflix Timed Text Style Guide: General Requirements). The BBC frames the same idea as 160–180 words per minute (BBC Subtitle Guidelines). Those numbers are your floor and ceiling for clip captions too.
The catch: podcast guests talk faster than scripted TV. A founder mid-rant or a comedian on a roll routinely blows past 17 CPS, and most auto-captioners just dump whatever was said onto the screen at whatever speed it came out. The result is a line that flashes and vanishes before a thumb-scrolling stranger can finish it. Below are the broadcast rules translated into numbers you can actually apply to a fast-talking clip, plus a 10-second self-check for any line you're unsure about.
What's the maximum reading speed for captions?
The working ceiling is about 17 characters per second for general adult content, dropping to roughly 13 CPS for kids' content (Netflix English (USA) Timed Text Style Guide). The BBC expresses it as 160–180 words per minute, which works out to around 15 CPS in English (BBC Subtitle Guidelines). These aren't arbitrary, they come from decades of broadcast captioning research on how fast a viewer can read and still follow the picture.
Why this matters more for a clip than for a film: on social, captions aren't an accessibility add-on, they're the whole experience. A widely repeated estimate puts around 85% of social video watched with the sound off (Digiday, 2016 publisher-reported data), treat it as directional, since studies range from roughly 69% to 85% and the figure is a decade old. The exact percentage moves; the direction holds. Most people meeting your show read it. If they can't read it in time, they scroll.
And clips are often the front door to the whole show, one studio's client data puts clips at 20–40% of new audience, with 2–5× reach lifts for video shows (Podcast Studio Glasgow), a single source's directional range rather than a platform audit. A caption running too fast to read is a front door that won't open for the silent majority.
The broadcast standards, translated for podcast clips
Here's the part nobody spells out: broadcast standards assume a 16:9 TV screen, a scripted pace, and a viewer who is sitting still. A vertical podcast clip is none of those. So the standards need translating, not copying. Below is the conversion I use.
| Parameter | Broadcast standard | Podcast-clip rule |
|---|---|---|
| Reading speed | 17 CPS adult / 15 CPS (BBC ~180 WPM) | Aim for 13–15 CPS, phones, distraction, on the move |
| Line length | ~42 characters | 30–38 characters on 9:16; big text fills the width fast |
| Lines on screen | 2 maximum | 1–2 lines; prefer 1 if the font is large |
| Minimum duration | 5/6 of a second | ~1 second minimum per cue |
| Maximum duration | 7 seconds | Split anything long; no static block of text |
Sources for the left column: Netflix English (USA) Timed Text Style Guide (17 CPS, 42 characters), Netflix General Requirements (two-line max, 5/6 sec minimum, 7 sec maximum), and the BBC Subtitle Guidelines (160–180 WPM). The right column is my practical adaptation for vertical clips.
Two translations need defending. Why aim lower than 17 CPS? Because the broadcast number assumes ideal conditions. A phone screen is small, the viewer is half-distracted, often walking, and the clip auto-plays mid-scroll with no warning. Shaving the ceiling to 13–15 CPS buys back the margin those conditions eat. Why shorter lines than 42 characters? Because a TV is wide and a 9:16 frame is narrow. At a caption size you can actually read on a phone, ~42 characters wraps awkwardly or shrinks the text. Thirty to thirty-eight characters fits one comfortable line at a legible size.
How to actually hit these numbers
You don't compute CPS by hand for every line. You set up the caption so the numbers fall into place, then spot-check the few that worry you.
- Pick a line-length cap, not a font size. Set your caption block to wrap at roughly 30–38 characters (or its visual equivalent, about 68% of the frame width, the same ratio the BBC uses for online video). This single setting prevents the most common failure: a line so long it has to shrink to fit, then becomes unreadable.
- Cap on-screen text at two lines. If a cue would need a third line, it's too much text for the time, split it into two cues. This is the rule broadcasters never break, and it's the easiest one to enforce.
- Hold every cue at least ~1 second. A flashing one-frame word is unreadable even if it's short. If a quick "yeah" lands on its own cue for a third of a second, merge it into the neighboring line.
- Let fast talkers run shorter, denser cues, don't cram. When the guest sprints, the temptation is to pack more words per line to keep up. Do the opposite: shorter lines, more cuts, so each one clears the reading-speed budget. The captions will lag the audio by a hair, which no one notices, versus flashing too fast, which everyone feels.
- Consider word-by-word for the fastest stretches. Karaoke-style captions that reveal one word at a time sidestep the line-length problem entirely, because there's never a full line to read at once. They're not free, see when word-by-word animated captions earn their keep, but for a rapid-fire monologue they keep the eye on pace.
If you're starting from scratch, the end-to-end captioning workflow covers transcribe → style → sync → burn-in. This article is the styling layer: it's what "good" looks like once the words are on screen.
The 10-second self-check for any line
When a caption feels too fast, you don't need software to confirm it. Count the characters in the line, count how many seconds it's on screen, and divide. Above ~17, it's too fast for broadcast; above ~15, too fast for a phone. That's the whole test.
Worked example: a 40-character line on screen for 2 seconds runs at 20 CPS, over the limit. You have three fixes, in order of preference. Split it into two cues (best). Trim filler, "you know," "I mean," "kind of", which buys characters without changing meaning. Or hold the line ~0.7 seconds longer if the pause in the audio allows. Run this on three or four lines per clip and you'll quickly learn whether your default settings are already safe or whether your guest needs tighter cutting.
Common mistakes that wreck reading speed
Letting auto-captions decide your line length. Most auto-captioners break lines on the pause in speech, not on a character count. A fast talker with no pauses gets one giant line that shrinks to fit. Set an explicit line-length cap instead of trusting the auto-break.
Centering big text and then writing long lines. Large captions look great until a 45-character line forces the font down two sizes to fit the frame. Decide the size you want first, then cap the line length to whatever fits comfortably at that size, usually 30–38 characters.
Three lines on screen. This is the one broadcast rule with no exception, and it's the easiest to violate when you paste a full sentence. Two lines, maximum. A third line means it's too much text for the cue's duration.
Flashing single words for a fraction of a second. Word-by-word captions are good; word-by-word where each word appears for three frames is unreadable. Set a minimum on-screen time of ~1 second per cue (or per word group) so nothing flickers past.
Polishing speech into clean prose. Rewriting "yeah, so the thing is" into "The key point is" desyncs the caption from the audio and reads as fake to anyone listening, and it doesn't save reading time, since you're adding characters as often as cutting them. Trim filler, keep the voice. The accuracy-versus-cleanup balance is its own decision; see how to fix caption accuracy without over-editing.
Tools and the honest tradeoff
You can enforce these numbers in a free editor by hand, set the line wrap, cap two lines, and eyeball the timing, but on a weekly show that's the step that eats your evening. AI captioners handle the wrapping and timing automatically, then leave you a review pass for the lines that run hot. The honest version: auto-first, then check reading speed on the few dense lines, exactly the same logic as auto versus manual captions overall.
No tool reads reading speed perfectly for you, because "too fast" is partly a judgment about how distracting the background is and how familiar the words are. Treat 13–15 CPS as the default and tighten further on busy footage or jargon-heavy lines. The font you choose also changes how much fits per line and how legible it stays at speed, the right caption font for small vertical screens gives you a few extra characters of headroom before text shrinks. And if you're posting natively, burn the captions in so your reading-speed choices actually ship; the burned-in versus soft captions tradeoff explains why hardcoded is the standard answer for clips.
FAQ
What is the maximum reading speed for subtitles? About 17 characters per second for adult content and 13 for children's, per the Netflix English (USA) Timed Text Style Guide; the BBC uses 160–180 words per minute, roughly 15 CPS. For vertical clips watched on phones, aim a notch lower, 13–15 CPS, to absorb distraction and small screens.
How many characters should a caption line be? Broadcast caps lines at about 42 characters, but that assumes a wide 16:9 screen. On a 9:16 vertical clip at a font you can read on a phone, 30–38 characters per line is the practical target, with one or two lines on screen at a time and never three.
How long should a caption stay on screen? At least five-sixths of a second per cue by broadcast standard (Netflix); round up to about one second for clips so nothing flickers. No single cue should run past seven seconds, split a long line into two so the reader gets fresh text instead of a static block.
My podcast guest talks really fast, how do I caption that? Don't pack more words per line to keep up. Use shorter, denser cues with more frequent cuts so each line clears the ~15 CPS budget, let the captions lag the audio by a hair (unnoticeable), and consider word-by-word reveals for the fastest stretches so there's never a full line to read at once.
How do I check if my captions are too fast? Count the characters in a line, count the seconds it's on screen, and divide. Over ~17 is too fast for broadcast; over ~15 is too fast for a phone. Fix it by splitting the line, trimming filler words, or holding the cue slightly longer if the audio allows.