Subtitle reading speed calculator

A mistimed subtitle fails in two opposite ways, and both are common in vertical video: if the text stays on screen too briefly relative to its length, no one finishes reading it before it changes; if it lingers too long for the little it says, reading feels slow and the person loses interest or skips ahead. The metric the professional subtitling industry uses to measure this is CPS (characters per second): how many characters of that line the viewer has to process for each second the subtitle is visible. It's a simple calculation —text length divided by time on screen— but very sensitive: the same sentence, with a third of a second less duration, can go from perfectly readable to impossible to follow.

In short-form, this problem gets worse for two reasons specific to the medium. First, a large share of Reels, TikToks and Shorts consumption happens with the audio off (on public transport, in the office, in shared spaces), which makes the subtitle the main channel —sometimes the only one— through which the message arrives; if that subtitle can't be read in time, the video simply isn't understood. Second, the editing pace of this format tends to be fast, with cuts every one or two seconds, leaving very tight time windows for each line of text. A script that sounds fine read aloud can produce subtitles with a very high CPS as soon as fast editing cuts are added.

There's a second, less-discussed but equally frequent problem: translations or adaptations from one language to another usually change the length of the text without changing the time available. A sentence in English that fits perfectly in the original duration may need considerably more characters in Spanish, and that translated subtitle ends up with a much higher CPS than the original without anyone planning it that way. Reviewing CPS line by line after translating or adapting a script is the only way to catch this before publishing.

• A 42-character line that stays on screen for barely 1.2 seconds gives a CPS of 35, well above readable: even if the script sounds natural read aloud, on screen it's impossible to follow for someone just starting to read that sentence.
• A short 4-word subtitle that stays 6 seconds on screen (because it got stuck to a long shot with no cut) feels artificially slow and creates the sense that the video "isn't moving", even though the CPS itself isn't a problem.
• When translating a script originally recorded in English, a sentence that had 30 characters in the original grows to 48 in Spanish while keeping the same time on screen — the CPS climbs from a comfortable range to a risky one without the timing being touched.
• A video with subtitles auto-generated by a transcription tool splits lines by audio pauses, not by reading length: some resulting lines end up with a very high CPS (lots of text, little time) while others are needlessly slow.