Split long subtitle lines

What the tool does to a cue

It joins the words of a cue, breaks them into lines no longer than your limit, and groups the lines into cues of at most your maximum number of lines. If everything fits in one cue it is simply re-wrapped. If not, the cue is split. This long one, with a limit of 42 characters and 2 lines:

1
00:00:01,000 --> 00:00:09,000
Welcome back to the channel, today we are going to look at how subtitle files are timed and why long lines are hard to read.

becomes two cues, and the eight seconds are shared between them according to their length.

1
00:00:01,000 --> 00:00:06,268
Welcome back to the channel, today we are
going to look at how subtitle files are

2
00:00:06,268 --> 00:00:09,000
timed and why long lines are hard to read.

The cue is divided at a word boundary, each piece is balanced into lines of similar length, and the first piece ends at the instant the second one begins. The first piece holds about two thirds of the characters, so it gets about two thirds of the time.

Why 42 characters per line, two lines

Subtitles compete with the picture for attention, so every convention is a compromise between how much a viewer can read and how much of the frame the text covers. A line of about 40 characters is short enough to read without moving your eyes far, and two lines is the most that fits under the picture without crowding it. Many English-language style guides land on a limit in the high 30s to low 40s, and 42 is the figure you will see most. Reading speed matters as much as line length: guides commonly also cap how many characters a viewer has to read per second, in roughly the high teens to around 20, which is why a long cue on screen for one second needs splitting or trimming and not only wrapping. These are conventions, so set the numbers your platform asks for. Languages with wider characters, or right-to-left scripts, often use different limits.

How the splitting is timed

When one cue becomes several, the duration is divided by character count. A piece with 60 characters gets twice the time of a piece with 30. This is an approximation of speaking time, since people do not say every character at the same speed, but it is far better than splitting the time equally, which flashes a short last fragment as long as a full line. If you have word-level timing from a transcript, that is better still; this tool works from the subtitle file alone. If a split cue looks slightly early or late, nudge the file with the time shifter.

Options explained

• Balance the lines narrows the line width as far as it can without adding a line, so a two-line cue is two similar lines instead of one full line and a short stub. Turn it off for the plain left-packed wrap.
• Leave cues alone when their own breaks already fit keeps hand-made line breaks, including dialogue with a hyphen for each speaker.
• Tags such as <i> do not count towards the line length, and if a line break falls inside italics the tool closes the tag at the end of the line and reopens it on the next so each line is valid on its own.
• Output format can be changed while you are there: paste SRT and download VTT, or the other way round.

Where subtitle text comes from

Machine-made captions are the usual source of overlong lines, because a transcript is written for reading as a document and not for fitting under a picture. If you are starting from a recording, make an SRT from the audio first, then wrap it here. The step-by-step process is in how to make SRT subtitles from a video, and transcript file formats explained covers how the formats differ.