How to Time Captions to Speech (2026)

Timing captions to speech means the words show up when they are said, and leave when a person could have read them. A character count does not do that. Neither does shifting the whole file by 200 milliseconds because one cue felt late.

How long a line may be is how many characters per line in subtitles. Which buttons move a cue is how to edit SRT timings. This page is the judgment.

Direct answer

1. Play the cue with the video, not from the script.
2. Start the cue as the first word starts. A little early feels like a spoiler. A little late looks broken.
3. End it after the phrase is readable, not in the middle of the next sentence.
4. Split a cue that holds two thoughts. Merge a flash that nobody can read.
5. On a short vertical video, do not let the cue outlive the clip.

What "in sync" actually is

People "fix sync" on a file whose open was already right. They then break the open. Hear cue one before you touch cue forty.

Start on the word

Park the playhead on the first sound of the phrase, not on the breath before it, and not after the first syllable. Set the cue start there.

A music hit is not a word. If the caption is for speech, do not pop the line on the beat unless the words start on the beat. Song lyrics are a different job, and this guide is not karaoke.

Names should appear when they are said. Putting the guest's name up during the host's introduction, because you know who is about to talk, is early. Wait for the name.

End when the line can be read

Speech is often faster than reading. A cue that dies the instant the speaker stops can be correct to the audio and useless to a person.

Leave it up long enough to finish the line, then take it down before the next phrase starts. If there is not enough gap, the line is too long. Cut words or split into two cues that each have a moment.

House warnings in the File Transcribe editor, which are a light and not a law:

• About 42 characters per line, fewer on a phone
• Two lines
• About 17 characters per second
• Shorter than half a second, or longer than about 7 seconds, gets a mark

An amber row means look at it. A 15-second caption of one slow sentence can be fine. A 15-second caption of four sentences is not.

Do not cover the cut

A cue that starts in one shot and ends in the next looks like a mistake even when the words are continuous. End the cue at the cut. Start the remainder on the next shot, if the sentence actually continues.

On a Reel or a Short, the whole video may be one shot and still be 12 seconds. A cue that hangs into the next clip in a sequence is the version of this bug people notice. Those files start at Instagram Reels when the source is a Reel. Time the captions to the cut you posted, not to a longer camera roll.

Silence, overlap, and laughter

Silence gets no caption, unless you are writing a sound description such as `[laughter]` and that description is part of the job. A talking-head Short usually does not need it. A course video sometimes does, when the laugh is the point.

Two people talking at once do not become one neat cue. Put the line you want a viewer to read. The other voice is a miss, not a timing problem. Moving the cue will not separate them.

A false start ("I think, wait, no") can stay or go. If you delete it, close the gap so the next cue does not sit on the dead air you just removed from the text while leaving it in the time.

A pass that takes one short video

1. Play the first cue. Move the start onto the first word.
2. Read the line aloud at a normal pace while it is on screen. If you finish after it vanishes, extend it or cut the line.
3. If the next sentence has already started, split.
4. Jump to a cut or a music change. Make sure a cue is not riding across it.
5. Export SRT or VTT and watch that same minute in the player you will publish to.

In Subtitles, type the start and end or drag the cue. The button detail is the other guide. Export is on Pro, about $19/month, and Plus, about $49/month. Free is about 50 minutes a month, TXT or PDF. Confirm on /pricing.

YouTube's panel can fix one late line on a video that is already up. It is a slow place to rebuild a whole short. Studio steps for a Short are how to add an SRT to YouTube Shorts.

FAQ

How do I time captions to speech?

Start each cue on the first word of the phrase and end it when a viewer can finish reading, before the next phrase. Check it against the video, not against the script alone.

Should captions match the audio exactly in time?

The start should. The end should prefer the reader when the speaker is faster than reading. A cue that vanishes on the last consonant is often too short.

What if the whole file is late?

If the first cue is late by the same amount as the last, shift the file once and listen to the open again. If only a later section is late, edit that section.

How long should a caption stay on a Short?

Long enough to read, and not longer than the shot. A line that is still up after the cut to the next clip is late even if the sentence was accurate.

Is reading speed the same as characters per line?

No. Characters per line is how wide the cue is. Reading speed is whether the time you gave that width is enough. A short line can still flash. A legal-length line can still sit there too long.

Open the video in Subtitles on File Transcribe, move the cue you can hear is wrong, and export when the plan includes SRT or VTT.

More guides

• Transcription guides
• Try File Transcribe free
• Transcript format guide
• Best transcription software
• Transcribe YouTube videos

Further reading

• W3C captions overview
• Google Speech-to-Text docs