Subtitle reading speed: CPS, WPM and the limits that matter

People read subtitles far slower than they read a book: they're also watching the picture, following the audio, and getting only a brief window per line. If text appears faster than it can be read, viewers miss either the words or the action. Reading speed is how we measure and prevent that.

Two ways to measure: CPS and WPM

• Characters per second (CPS) — the modern standard. Count the characters in a cue (including spaces, excluding markup), divide by the cue's on-screen duration in seconds. A 40-character line shown for 2 seconds is 20 CPS.
• Words per minute (WPM) — the older, broadcast-era measure. Roughly, WPM ≈ CPS × 10–11, so 17 CPS is about 180 WPM. CPS is now preferred because it doesn't depend on how long words happen to be.

The numbers the industry uses

Standard	Reading-speed limit	Notes
Netflix — adult programmes	17 CPS	The most-cited modern target
Netflix — children's	13 CPS	Younger viewers read slower
BBC subtitle guidelines	~160–180 WPM	≈ 15–17 CPS, broadcast captioning
Common tool ceiling	~20 CPS	An upper readability limit; above it, flag for review

Think of it as a target and a ceiling. 17 CPS is a good aim for comfortable reading; 20 CPS is roughly where most viewers start to lose either the text or the image. Our SRT validator flags any cue above 20 CPS so you can review it — deliberately a ceiling, not the stricter style-guide target, so it catches genuine problems without nagging about every line.

The other constraints that work with it

Reading speed doesn't act alone. The companion rules from the major style guides:

• Maximum line length — around 42 characters per line (Netflix); BBC uses a percentage of screen width. Longer lines wrap awkwardly or run off-screen.
• Maximum two lines per cue — a third line covers too much picture and is hard to scan.
• Minimum duration — about 5⁄6 of a second (≈0.83 s), even for a single word, so the eye can register it.
• Maximum duration — around 7 seconds; beyond that a static line invites re-reading.
• Minimum gap — a couple of frames between consecutive cues so viewers notice the change.

These interact: if a line is over the CPS limit, your options are to extend its duration (if there's room before the next cue), shorten the text (condense without losing meaning), or split it across two cues. Which one is right is an editorial judgement — which is exactly why a good tool flags the cue but never rewrites it for you.

Why we flag, but never auto-fix, reading speed

Every other structural fault has a correct mechanical answer: a missing blank line gets added, an overlap gets trimmed, a backwards timestamp gets repaired. Reading speed is different. Shortening a line means choosing which words to cut; extending a cue might bleed it over a scene change. There is no safe automatic choice, so our repair tool lists the offending cues and leaves them intact. You make the call in the editor, where you can see each cue's text and timing together.

The accessibility dimension

Reading speed isn't only a quality nicety — it's central to accessibility. Captions exist so that d/Deaf and hard-of-hearing viewers, and anyone watching without sound, get equal access to the content. A caption that's technically present but unreadably fast doesn't deliver that. Captioning quality frameworks treat appropriate reading speed as a core requirement, alongside accuracy, synchronisation and completeness. When you respect CPS limits, you're not just polishing — you're making the video usable for more people.

A practical checklist

1. Run the file through the validator and note the cues flagged over 20 CPS.
2. For each, decide: extend the timing, condense the wording, or split into two cues.
3. Aim for ≤17 CPS where you can, ≤13 for children's content.
4. Keep lines under ~42 characters and cues to two lines.
5. Make the edits in the editor and re-check.

Get reading speed right and the rest of the craft — timing, line breaks, accuracy — has room to land. Get it wrong and even a perfect translation never gets read.