Find Too-Long Subtitles
Or paste subtitle text
Options
A duration — 7000, 7s and 00:00:07,000 all work. The other end of the line from the shortest-time box: a subtitle still on screen after this has usually outlived its line, and viewers read it a second time or start waiting for it to go. This box belongs to the check pages, so the file comes back untouched — the report lists every cue over the line with its measured duration next to the limit. It and the shortest-time box are two ends of one question, so a page carries one of them and not both.
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.
A subtitle that stays too long is a line that outlives its shot. It sits on screen over the next cut, over the next speaker, sometimes over a whole scene, telling the viewer something that is no longer true. This page lists every cue longer than the maximum you set, with its timecodes, its actual duration and the limit it was measured against — as a report, leaving the file untouched.
Long cues are the quieter half of the duration problem, and the more common one in material that has been through several hands. A line held across a scene change is what a viewer notices first, because the next shot's dialogue arrives underneath it. The usual causes are mechanical: a timing pass that stretched a cue to the end of the shot instead of the end of the sentence, a merge of two cues where only one end was extended, a translation that grew longer than the line it replaced while the timing stayed as it was. The report separates those cases by showing how far past the limit each cue goes — a cue seven seconds past a seven-second ceiling and one four hundred milliseconds past it are different problems.
The ceiling is inclusive, the same reading as on the too-short page: a cue is reported only when it is longer than the number, so the number you type is the longest a cue may be. The two pages share a rule and ask opposite questions, which is why each declares one threshold and not both — set both at once and a report cannot tell you which side of the range you are reading about. LRC and plain text are not accepted, since neither holds an end time to measure a duration from.
Examples
A line that outlives its shot
Subtitles
1
00:00:01,000 --> 00:00:03,000
Where were you?
2
00:00:03,000 --> 00:00:12,000
I was working late, and then I stayed on to finish the report, and by the time I looked up it was morning.
What the report finds
Cue duration outside the limit — 1 found
Severity Cues Start End Measured Limit
Warning 2 00:00:03.000 00:00:12.000 9000 ms 7000 ms
Nine seconds against a ceiling of seven. The file is not touched: cutting a long cue means deciding which half of the sentence can go, and that is a person's job rather than a timing tool's, which is exactly why this page reports instead of repairing.
Exactly at the ceiling
Subtitles
1
00:00:01,000 --> 00:00:08,000
I was working late, and then I stayed on to finish the report.
What the report finds
Cue duration outside the limit — Nothing found
Cues 1
Checks run 1
Seven seconds exactly against a ceiling of seven, and nothing is reported. The limit is the longest a cue may be, so it is inclusive in the same direction as the minimum on the too-short page: the number you type is legal, and only what goes past it is a finding.
A cue held across a scene
Subtitles
1
00:00:03,000 --> 00:01:03,000
Hold on.
2
00:01:04,000 --> 00:01:05,000
What?
What the report finds
Cue duration outside the limit — 1 found
Severity Cues Start End Measured Limit
Warning 1 00:00:03.000 00:01:03.000 60000 ms 7000 ms
A minute-long cue, and the one-second cue after it correctly left alone. A cue held for a whole scene is the most visible form of this defect — the viewer reads the line, then reads it again over the next shot's dialogue — and it also swallows every gap and overlap check that runs on the same span.
Frequently asked questions
When is a subtitle too long on this page?
When it stays on screen for longer than the maximum you set, in milliseconds. The comparison is strictly longer than, so a cue lasting exactly the maximum is not reported. The maximum applies to the whole run and is printed in the limit column of every finding it produces.
What maximum duration should I use?
Seven seconds is the default and the number most style guides land near. It is not a rule this page is applying — it is a starting point you can change, and the report always shows the value you actually ran with, so a report read later does not leave the reader guessing what was measured.
How long a cue is too long for reading, rather than for timing?
That is a reading-speed question and a different page: a cue can be short in seconds and still contain more words than anyone can read in the time. This page measures duration only, against the number you set, and it makes no attempt to decide whether the text fits.
Does this page split or shorten long cues?
No. It reports them and nothing else — no timecode is moved and no text is cut. Splitting a cue in two changes what the viewer reads and when, which is a decision that belongs to whoever knows the material; this page's contribution is telling you which cues need that decision made.
Why does a cue end up too long?
Usually a timing pass that stretched the cue to the end of the shot rather than the end of the sentence, or a merge of two adjacent cues where only one of the ends was extended. Translation is the other common source: the translated line is longer than the one it replaced and the timing was never revisited, so the cue stays up long after the reading is done.
Which subtitle formats can be checked for long cues?
The ten that store both a start and an end time: SRT, WebVTT, ASS, SSA, SBV, SMI, MicroDVD, TTML, JSON and CSV. LRC and plain text are not accepted, because a duration needs two times and neither format has them.
Can this and the short-cue check run on the same file?
They can, but not on the same page — each page declares one threshold, so a run produces either the short list or the long list. Running both over one file means two passes, and that is deliberate: a single report with two thresholds cannot say which side of the range a finding belongs to without making the reader check the numbers.