Merge Subtitle Files

"Merge subtitles" means two entirely different jobs, and the tool that does one badly is usually the tool that thought there was only one. Both are here, because the arithmetic is different and so is what you lose.

Two languages at once

The bilingual case: the same film, two translations, and both should be on screen together — for language learning, for a mixed audience, or for a review pass where the translator wants the source line under their own. Dual subtitles are almost never available as a download, so joining two tracks is how anyone gets them.

The first track’s timing wins, and that is a real loss. Two translations of the same scene almost never break their lines at the same moment — a sentence that needs one cue in English needs two in Arabic, and the pauses fall elsewhere. One set of cut points has to be chosen, so the language you load first is the one that stays in step with the speech, and the second is fitted to it.

A cue in the second track that has nothing on screen at the same moment is kept as a cue of its own rather than dropped. It is still a line somebody wrote, and losing it would be a silent deletion — nothing on screen would reveal it.

The second half of a film

The other case: a film that arrived in two parts, with two subtitle files, and the second one starting its clock again from zero. Combine them naively and every line of part two lands in the first ninety seconds of the film, on top of the opening scene.

The offset is yours to give. It is the running time of part one, and it has to be read off the video file rather than inferred from the subtitles — because the last line of dialogue in part one is not the end of part one. A tool that guesses from the final cue is short by the length of the closing scene, every time.

Press Load a sample. Two tracks appear: two English cues, and their Arabic translation. Look at the timings before you merge — the Arabic cue starts at 00:00:01,200 and ends at 00:00:03,800, while its English partner runs from 00:00:01,000 to 00:00:04,000. Two hundred milliseconds apart at each end, which is what two independent timing passes look like.

Merge them with two languages at once. The first cue comes out as 00:00:01,000 --> 00:00:04,000 with both lines under it. The Arabic timing is gone — not averaged, not split the difference, gone — and the status line reads paired 2 so you can see it happened to both cues rather than to one.

That is the trade the mode makes. If the Arabic timing is the better of the two, load Arabic first and English second, and it is the English that gets fitted.

Now switch to the second half of a film and set the offset to 01:02:30. The same second file is now appended an hour, two minutes and thirty seconds later, and nothing is stacked at all — every cue keeps its own line and the two tracks simply run one after the other.

If the halves came from different releases, the join will be right and the sync may not be. Retime the second file first, then merge. And if the stacked track reads too densely — two languages is twice the text in the same seconds — check the reading speed before you call it finished.