Split Long Subtitles
Or paste subtitle text
Options
Counted in characters as a reader sees them rather than in bytes or UTF-16 units, so an accented letter counts as one. Any cue holding more than this is cut into pieces of at most this many characters. The cut falls on a space, at the end of a line, or between Chinese and Japanese characters; only a single unbroken run with none of those, such as a long URL, is cut mid-word. Each piece keeps a share of the cue's time in proportion to how much text it carries.
Which frame rate the timings are counted in. On the format pages this is what MicroDVD needs, because it stores timing as frame numbers. On the frame rate pages it is the rate the file was timed against — the version the subtitles were made for — and the page's second box takes the rate of the video you are going to use them with. 23.976, 24, 25, 29.97, 30, 50, 59.94 and 60 are the common ones. Getting it wrong shifts the whole timeline, and on the frame rate pages it shifts it by a ratio rather than by a fixed amount.
This page cuts any cue that holds more text than your limit into pieces that each fit. The limit is counted in characters as a reader sees them, and it starts at 42 — the per-line ceiling that subtitle style guides have converged on, and the number a broadcaster or a streaming platform will be checking your file against. A cue holding 60 characters comes out as two cues; the text is divided at the nearest space, and every character that went in comes out again.
The pieces share the original cue's time in proportion to how much text each one carries. That is the second half of the same idea: a long line is not only too much to read at once, it also needs longer on screen, and reading speed is what the limit is really about. The span as a whole does not move — the first piece starts where the cue started, the last one ends where it ended, and nothing else in the file shifts by a millisecond.
The cut lands on a space, at the end of a line, or beside a Chinese or Japanese character. Only a single unbroken run with none of those — a long URL, a string of digits — is cut mid-word, because the limit is a hard ceiling and there is nothing on the panel that could express "this one could not be divided". Scripts with neither spaces nor ideographic characters, such as Thai or Khmer, fall into the same case.
One thing to know before you run it: splitting changes how many cues the file holds, and a file whose cue count changed is written out in timeline order with fresh numbering from 1. If any cue carries a style or a position — an ASS file, a styled WebVTT — that information stays with the first piece, and the pieces the tool creates have none; the report counts them so you know how many are affected. Everything runs in this browser tab: no file is uploaded, and SRT, WebVTT, ASS, SSA, SBV, SAMI, MicroDVD, TTML, JSON and CSV are all read the same way.
Examples
A 56-character line with the limit at 42
Input
1
00:00:01,000 --> 00:00:04,000
She walked into the room and closed the door behind her.
Output
1
00:00:01,000 --> 00:00:03,127
She walked into the room and closed the
2
00:00:03,127 --> 00:00:04,000
door behind her.
Two pieces of 39 and 16 characters, and the three seconds are shared 2127 to 873. The cut is on the space before door rather than exactly at the 42nd character, so no word is broken. If this cue had a style, the first piece would keep it and the report would say one new cue came out unstyled.
Chinese text, where there are no spaces to break on
Input
1
00:00:01,000 --> 00:00:04,000
我们今天下午三点在会议室开会讨论明年的计划。
Output
1
00:00:01,000 --> 00:00:02,364
我们今天下午三点在会
2
00:00:02,364 --> 00:00:03,727
议室开会讨论明年的计
3
00:00:03,727 --> 00:00:04,000
划。
With a limit of 10, the 22 characters come out as 10, 10 and 2. A break beside an ideographic character counts as a break point, so Chinese and Japanese files divide on their own terms; each character also counts as one character, so the limit means the same thing here as it does in English.
A file that is already inside the limit
Input
1
00:00:01,000 --> 00:00:02,000
A short one.
2
00:00:03,000 --> 00:00:04,000
And another.
Output
1
00:00:01,000 --> 00:00:02,000
A short one.
2
00:00:03,000 --> 00:00:04,000
And another.
Nothing here is longer than 100 characters, so nothing is divided and the output is the input byte for byte, numbers included. This is the run to make first when you are not sure what is in a file: a limit set too high can only do nothing.
Frequently asked questions
Where does the 42-character limit on this page come from?
It is the per-line ceiling that most subtitle style guides and broadcaster specifications settle on, and it comes from reading behaviour rather than from any one document: a viewer tracks about 42 characters comfortably in a single line at normal viewing distance. Two such lines are the usual maximum for one cue, which is why 42 is where this box starts. Raise it for a file you are only reading on a computer, and lower it for material meant to be watched on a phone at arm's length.
Will splitting a line change when the subtitle appears on screen?
The cue as a whole appears and disappears at exactly the same moments: the first piece starts where the original started, the last ends where the original ended. Inside that span the pieces take turns, in proportion to how much text each one holds, so a piece with twice the text stays up roughly twice as long. That proportional share is worked out from the character counts, not from anything in the file, since the file only ever said when the whole line was on screen.
Does the tool renumber the file after splitting a cue?
Yes, and it is worth knowing why. A file whose cue count changed is written in timeline order and numbered from 1, so the numbers stay a count of what is in the file rather than a set of labels with gaps in it. The practical consequence is that a file you split can no longer be compared to the original cue by cue on the number line — compare them on the timeline preview instead, or keep the original next to you.
Can splitting lose text or break a styled cue?
No character is dropped: every piece is a continuous run of the original text, including any line breaks inside it, and only the whitespace at the two cut edges is trimmed away. What changes is that a style, a position or a karaoke timing belongs to the cue it was written on, so it stays with the first piece and the pieces created after it have none. The report counts the cues that lost their formatting this way, and you can see it before downloading.
What happens to a word longer than the limit?
It gets cut through the middle, which is the one case where the tool breaks a word. The alternative would be to let the piece run past the limit, and the limit is what you set it for, so the cut wins. Long URLs, hashes and strings of digits are the usual culprits. Raise the limit above the longest unbroken run in the file if you would rather keep those whole, and remember that Thai or Khmer text behaves like a long URL here — it has no spaces and no ideographic characters to break beside.
Should I use this page or the one that splits by duration?
Ask what the reader is failing at. If a line is on screen for long enough but holds too much to read in one go, the problem is characters and this is the page. If a line holds a comfortable amount of text but lingers on screen after everyone has read it, the problem is seconds and the duration page is the one. A cue that is both too long and too slow needs both runs, and running this one first is the safer order: it can only shorten the pieces, while raising or lowering a duration limit changes timings you may have already set.