Subtitle Line Length Checker

Two limits, and they are not two versions of the same limit. One is about the screen and one is about the eye, and a cue can pass either while failing the other.

Characters per line — CPL — is about the frame. Past roughly 42 the text either runs to the edges or the player wraps it for you, and a player’s wrap is purely mechanical: it breaks at whatever word runs out of room, not where the sense breaks. A line that reads "he told her he had never once considered leaving" becomes two lines split after never, and the viewer reads a clause that means the opposite for the third of a second before the second line registers.

Lines per cue is about time. A third line is legal in most specifications and wrong in nearly all of them. The block climbs high enough to cover the picture, and three lines take longer to read than most shots stay on screen — so the viewer either misses the line or misses the shot. Two lines is the standard, and one is better when it fits.

Where 42 comes from, and why yours may differ

It is a subtitle standards figure rather than a law, and the ones in circulation disagree because they were written for different screens. 42 is Netflix’s limit for most Latin-script languages. 37 is the teletext figure the BBC and much of European broadcast still work to, because a teletext row was 40 characters wide and needed margins. 39 is a common house compromise. Cinema subtitles run wider than any of them. The limit is a setting here for that reason.

Markup is not counted

<i>whispering</i> is ten characters on screen and twenty-seven in the file. Counting the file rather than the screen flags lines that are perfectly fine and, worse, misses lines that are not — because a cue can be under the limit in characters and over it once the tags come off nothing. Every count here is made on the visible text, after the tags are removed, per line rather than for the cue as a whole.

What it does not do is rewrite anything. Where a line should break is a decision about where the sense breaks, and it belongs to whoever wrote the line.

Press Load a sample, and leave the limits at 42 and 2. Two of the four cues are listed, and they fail for opposite reasons.

Cue 2 is one line of 63 characters. One line — well inside the line limit — and half as wide again as the frame allows. It is at the top of the list because the list is sorted by the longest line rather than by cue order, which is the order you want to work through them in.

Cue 3 is "Longer / than you / would think." — three lines, the longest of them twelve characters. Nothing about it is too wide. It is on the list because three lines is a line too many, and this is the case a character count on its own never catches.

Cue 4 is two lines of 25 and 19 characters and does not appear at all. Cue 1 is nineteen characters on one line. Both are fine, and a checker that listed them would train you to stop reading the list.

Now set the width to 37. Cue 4’s longest line is 25, so it still passes — the teletext limit does not catch it, and neither does anything else, because it was written for two lines from the start.

Line length and reading speed are the two halves of the same question and they are genuinely different measurements: a 60-character line is fine over four seconds and impossible over one. Run the reading-speed check alongside this one. And if the lines are long because the file came out of Aegisub — where the renderer was free to shrink the font — the ASS converter is where that track came from.