Point Sync Subtitles
Or paste subtitle text
Options
A timecode you can see in the subtitle file right now: the cue that lands on a line of dialogue you can point at. Paired with the box under it, the two make one sync point — the tool moves everything so that moment lands where it should.
Where that same moment belongs — read it off the video. The difference between the two boxes is the offset applied to the whole file.
A second timecode from later in the file, at least a few minutes after the first. One sync point can only move the timeline; adding a second also stretches or squeezes it, which is what a file needs when it starts close to right and drifts steadily out towards the end. Leave both of the lower boxes empty to fix the offset alone.
Where the second moment belongs. With both points filled in the tool works out the ratio and the offset together, so the beginning and the end both land correctly. If the two points disagree about direction the ratio cannot be derived and only the first point's offset is applied.
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.
Fill in In the file, this moment to run.
Point at a moment you can see in both places — a line in the subtitle file, the same line in the video — and the whole timeline moves so that moment lands where it should. Fill the first pair of boxes and the file is offset. Fill the second pair as well, at least a few minutes further in, and the file is stretched or squeezed to fit both, which is what a file needs when it starts close to right and drifts steadily out towards the end.
One reference point can only move the timeline; two also change its speed. That is the whole difference, and it is the difference between a file that is uniformly late and a file that was timed against a slightly different version of the video. A uniform error is obvious — every cue is out by the same four seconds — and one pair of boxes fixes it. A growing error is not: the first minute looks right, five minutes in the subtitles are a quarter of a second behind the picture, and by twenty minutes they are a full second behind, and no single offset can be right at both ends. Two points solve for the ratio and the offset together, so both ends land on the numbers you gave.
Both pairs are read off clocks, so no arithmetic is asked of you. What the arithmetic needs is that the two moments be far apart and unambiguous: a sign reading, a name card, the first line after a scene change. Two points a few seconds apart carry almost no information about speed, and a pair that disagrees about which way the file is out cannot produce a ratio at all — when that happens only the first point's offset is applied, and the report says something was dropped so the disagreement is not silent.
The two pairs describe one straight line, so this is not a way to fix the first half of a file with one pair and the second half with another; a file that is right up to a point and wrong after it has been joined from two pieces, and wants splitting instead. Everything runs locally in this browser tab — SRT, WebVTT, ASS, SSA, SBV, SAMI, MicroDVD, TTML, JSON and CSV, single files or a whole season in one queue, with no upload at any stage.
Examples
One pair, because the whole file is out by the same amount
Input
1
00:00:05,000 --> 00:00:07,000
Hello there.
2
00:00:12,000 --> 00:00:14,000
Second line.
Output
1
00:00:04,000 --> 00:00:06,000
Hello there.
2
00:00:11,000 --> 00:00:13,000
Second line.
The file says this line is at five seconds, the video says four, so one second comes off every cue. That is the two-times-slower version of filling in an offset: nothing was worked out by hand, and the pair of boxes is the record of why the file moved.
Two pairs, because the error grows
Input
1
00:00:00,000 --> 00:00:02,000
First.
2
00:05:00,000 --> 00:05:02,000
Five minutes in.
3
00:20:00,000 --> 00:20:02,000
Twenty minutes in.
Output
1
00:00:00,000 --> 00:00:02,002
First.
2
00:05:00,250 --> 00:05:02,252
Five minutes in.
3
00:20:01,000 --> 00:20:03,002
Twenty minutes in.
The first cue is already right and the twenty-minute mark is a second late, with the reference points given as 00:00:00,000 → 00:00:00,000 and 00:20:00,000 → 00:20:01,000. Both anchors land exactly where they were asked to and the cue in between — at five minutes, which is a quarter of the way along — comes out a quarter of a second late, which is what a single ratio does to everything between the ends. Note that no pair of offset boxes could have produced this: any one amount would be right at one end and wrong at the other.
The second pair disagrees with the first
Input
1
00:00:01,000 --> 00:00:02,000
First.
2
00:00:03,000 --> 00:00:04,000
Second.
Output
1
00:00:02,000 --> 00:00:03,000
First.
2
00:00:04,000 --> 00:00:05,000
Second.
The first point says the file is a second early, the second says it is four seconds late. Taken at face value the two would require the timeline to run backwards, with every cue ending before it starts, so no stretch is derived and only the first point's offset is applied — everything moves one second later, exactly as if the lower pair had been left empty. The loss report counts one item, which is how you find out the two references disagree rather than getting a file that looks plausible.
A reference point in the middle of the file
Input
1
00:00:01,000 --> 00:00:02,000
Before the reference.
2
00:00:03,000 --> 00:00:06,000
Straddling it.
3
00:00:05,000 --> 00:00:07,000
On the reference.
Output
2
00:00:00,000 --> 00:00:02,000
Straddling it.
3
00:00:01,000 --> 00:00:03,000
On the reference.
The moment you point at does not have to be the file's first cue, and this is what that makes possible: the file says the line is at five seconds, the video says one, four seconds come off everything, and the first cue — which only started at one second — has nowhere left to go. A cue that ends before zero is dropped and counted; one that keeps its closing time is pinned to zero and comes out shorter; the cue on the reference lands exactly where it was asked to. Cue numbers stay as the file had them, so the two that survive read 2 and 3.
Frequently asked questions
How do I tell whether my file needs one pair of reference points or two?
Fill in the first pair, then check a moment ten minutes further in. If that moment lands where you expect, the file was uniformly out and the lower two boxes are not needed. If the start is right and the ten-minute mark is still behind the picture, the error is growing and the second pair is what fixes it — nothing else on this page will.
Which two moments make a good pair of reference points?
Two you can identify beyond doubt and far apart: a sign reading, a name card, the first line after a scene change, a word you can hear. Minutes apart, not seconds. Two points five seconds apart still give a ratio, but a rounding error in either timecode becomes a large error in that ratio, and the further apart they are the more the speed is pinned down.
Why did the tool ignore my second pair and just move the whole file?
Because the two points contradict each other. If the first says the file is early and the second says it is late, or both describe the same moment, no single line can pass through both, so only the first point's offset is applied and the report counts one item dropped. Check the direction of both pairs against the video: the file has to be out the same way at both of them.
Can I fix the first half of a file with one pair and the second half with another?
No. The two pairs describe one straight line, so both are already fixed by the same ratio and offset, and adding a third pair is not possible. A file that is correct up to a point and wrong after it has been joined from two pieces, or timed against a version assembled from reels running at different speeds: split it where the error starts and give each half its own page run.
My subtitles are simply half a second out from beginning to end. Do I need the second pair?
No, and filling it in would make things worse rather than better. A uniform error is exactly what one pair of boxes measures, and it is the safer of the two to work with: a single pair can only move the file, so it cannot introduce a speed error where there was none. Save the second pair for a file that is right at the start and wrong later.
What happens when the moment I point at is not the first cue in the file?
The offset is still applied to the whole file, so the cues before that moment move too — and if your reference point is early in the file, some of them can be pushed before the start of the timeline. Those are handled the same way negative offsets are: a cue that still has time left on screen is pinned to zero and comes out shorter, a cue with none is dropped, and both are counted in the report.