Subtitle line breaks: make fast captions easier to read

First decide whether the problem is width or time

A subtitle cue is one timed block of text. A line break changes how that block is arranged; it does not change how long the block stays on screen. If the words disappear before you can finish reading them, moving half the sentence onto a second line will not solve the time shortage. If the words fit comfortably in time but wrap into an awkward shape, start with the line break.

Work on a copy of a recording and subtitle file you own or are authorized to edit. Keep the original text and timing as a reference. Watch a difficult passage at normal speed and note three things: where your eyes hesitate, whether you finish before the cue disappears, and whether the caption hides something you need to see. Those observations lead to different fixes.

Break between phrases, not inside them

Try to keep words that belong together on the same line. The BBC subtitle guidelines prioritize meaningful phrase boundaries over a perfectly balanced shape. Netflix’s general requirements similarly discourage separating names, articles from nouns, or parts of a verb. These are useful editorial principles; the full delivery specifications belong to their respective services.

Consider this invented line from an authorized studio tour. Both versions contain the same words. The second keeps the name together and puts the complete instruction on the next line. It is easier to parse even though the two lines have different lengths. Prefer one line when it fits the destination comfortably; do not insert a break merely to fill a two-line template.

Awkward:
Ask Maya
Chen to open the blue folder.

Clearer:
Ask Maya Chen
to open the blue folder.

Keep a line break inside the same cue

In SRT, a new text line inside a cue is different from a blank line between cues. To arrange one caption over two lines, keep its number and timestamp line, then put the second text line immediately below the first. A blank separator belongs after the complete cue. The example below remains one cue that lasts three seconds.

Splitting that text into two separately timed cues is a different editing operation: it needs a suitable boundary in the speech and enough display time for each part. Do not invent a new start time just because the sentence looks long. Save a separate revision, and check the actual exported file instead of assuming the editor’s visual wrapping is a stored line break.

12
00:00:10,000 --> 00:00:13,000
Ask Maya Chen
to open the blue folder.

Use reading speed to find cues worth reviewing

A simple words-per-minute check is the word count divided by the cue duration in seconds, multiplied by 60. For this English example, count whitespace-separated words. The nine-word caption below lasts two seconds: 9 ÷ 2 × 60 = 270 words per minute. If the same words had three seconds, the result would be 180. This is an arithmetic comparison, not a recommendation to extend this particular cue.

The BBC recommends a range of 160–180 words per minute while treating timing as an editorial decision. Use that as a sourced reference, not a universal pass/fail rule for every language, audience or platform. A person reading an unfamiliar name, a figure or a technical term may also need to look at the picture. Review the cue in context even when its calculated rate looks comfortable.

If your delivery specification uses characters per second instead, follow its counting convention and threshold. Words per minute and characters per second are different measures. Do not copy a number from one into the other or assume a single English word-count rule works equally well across languages.

00:00:20,000 --> 00:00:22,000
Keep the original file before
you export another version.

9 / 2 × 60 = 270 WPM

Reduce pressure without changing the message

First correct recognition errors: an accidental repetition can make a cue both longer and less accurate. Then review punctuation and wording against the audio. For a translation, a fluent reviewer can look for a shorter natural equivalent that preserves the meaning and tone. Do not remove a qualification, negation, name or number just to make a counter look better. W3C warns that even a missing negative can reverse an automatic caption’s meaning.

If faithful text still cannot be read in time, flag the passage for a timing or segmentation review. Extending a cue may run into the next speaker, a scene change or important visual information. Splitting it may create two flashes that are harder to follow. A global timing shift moves cues earlier or later but does not increase their duration, so it cannot fix dense text. Resolve that choice with the recording in view.

Treat character limits as destination rules

A character count is not a measurement of rendered width. The BBC explains that font, font size and individual letter widths affect how a line fits. A string of wide letters can occupy more space than the same number of narrow letters. This is why a line that looks fine in a text editor may wrap differently in a video player.

Netflix’s general delivery guide sets a maximum of two lines and directs readers to language-specific requirements. That is a Netflix delivery rule, not a property of the SRT format or a guarantee of accessibility. Follow the requirements of your actual destination, then preview your file there. Check both the usual landscape layout and a narrow viewing window; use the player’s larger caption setting when available.

Review the text locally in Polyglot

Polyglot on Apple Silicon Macs can create subtitles from media or import an existing SRT or UTF-8 WebVTT file. It supports local text review and subtitle export. Make corrections against the recording, then review each translated language separately: a good break in the original sentence may land inside a phrase in the translation. Editing the original can mark affected translations for review; do not assume an already completed translation has been corrected for you.

This guide does not assume an automatic reading-speed checker or automatic phrase reflow in Polyglot. Use the calculations above as a manual review aid. If you need a precise stored line break, inspect the exported SRT and adjust a copy in a plain-text editor where necessary. Keep the saved project and final delivery revision clearly named so later exports do not accidentally replace the reviewed copy.

SRT and WebVTT exports are separate subtitle files. The receiving player controls their final appearance, and imported WebVTT presentation is flattened with a notice in Polyglot. Preview the exported result in the destination; a successful export alone cannot show that every line is readable.

Make one final pass at normal speed

Watch the final video with the exact subtitle revision you intend to deliver. Read without pausing through the densest passage, a name or number, a speaker change and a caption beside important on-screen text. Check that names stay together, the message remains faithful and no revised cue collides with its neighbor. For translated captions, have someone who understands the target language review natural phrasing.

Finally, watch with the sound off. Readability is only one part of caption quality: relevant sound information and speaker identity may also be needed, as W3C’s caption guidance explains. Clean two-line dialogue alone is not proof that a video meets an accessibility standard. Keep unresolved passages on a review list instead of deleting meaningful information to meet a formatting target.

Keep exploring.