The secrets of subtitle timing
Everyone notices bad subtitle timing. Nobody notices good timing. That asymmetry is the whole game — and it is why timing is the hardest part of subtitling.
This story is about how professional subtitlers think about timing. Not the guideline numbers, but the intuition behind them. The feel of a well-timed cue. The mistakes that make a subtitle feel "off" even when nothing is technically wrong.
Lead-in and lead-out
Every cue has a lead-in and a lead-out — the small amounts of time before the actor starts speaking and after they finish.
Lead-in gives the viewer's eye time to jump to the subtitle and start reading. The standard is 100–200ms. Too short and the cue feels like it slaps into place. Too long and the viewer reads ahead of the audio, which is confusing.
Lead-out gives the viewer time to finish reading. The standard is 200–400ms. Too short and the cue disappears mid-read. Too long and it hangs around into the next scene or the next line of dialogue.
The trick is that these values are not fixed. They scale with the cue's content. A dense, information-heavy cue needs more lead-out. A one-word cue can disappear almost immediately. A cue with unusual vocabulary (proper nouns, technical terms) needs more lead-in, because the eye takes longer to land.
Reading rhythm
A sequence of well-timed cues has a rhythm. The eye moves from one line to the next in a predictable pattern. The viewer does not have to consciously track the subtitles because the rhythm carries them.
When does a sequence have rhythm? Three conditions:
- Consistent lead-in. Every cue starts about the same amount of time before its line begins.
- Consistent lead-out. Every cue ends about the same amount of time after its line ends.
- Reasonable pacing. Cues don't cluster (10 cues in 5 seconds) or spread out (a cue per minute).
Most professional subtitlers aim for this without consciously thinking about it. It comes from years of experience, and from feeling the discomfort when a sequence does not feel right.
The problem of overlap
Dialogue in film is not polite. People talk over each other, murmur under music, pause mid-sentence. A cue that ends when the actor stops speaking might overlap with the next cue that starts when the next actor starts speaking. Now the viewer sees two lines of text at once, and only one of them is being spoken at a time.
The professional solution is either to split the cue into two shorter cues (so each one ends before the next starts) or to merge the two lines into a single cue with stacked dialogue. Neither is perfect. Both require judgment.
This is where subtitling becomes genuinely hard. The tool cannot make the decision for you. A new subtitler might merge everything into one cue, producing dense blocks of text. An experienced subtitler reads each moment and decides.
Time outside the audio
Some subtitles need to stay on screen when no one is speaking. A caption for a title card. A translation of a sign. An annotation for a sound effect. These cues are anchored to a visual, not to audio.
The rules for these cues are different. They need enough time to read, but they don't need to align with dialogue. A sign that appears for three seconds in the video should be subtitled for three seconds — not less, not more.
Forced narrative subtitles (translating foreign-language dialogue within a primarily monolingual film) are a special case: they need to convey meaning even if the foreign dialogue is not fully audible. The subtitler often has to over-describe or explain.
Timing for accessibility
SDH subtitles (for deaf and hard-of-hearing viewers) have additional timing considerations:
- Sound effects like
[door slams]need a cue even if no dialogue is happening. This cue must appear at the moment of the sound and stay long enough to read. - Speaker labels add to reading time. A cue that would be borderline for general audiences might need to be split when it has speaker identification.
- Music and lyrics need their own cue. If the lyrics are sung over dialogue, the SDH cue must coordinate with the dialogue cue.
All of this eats into the tight timing budget of a scene. Good SDH subtitlers often have to compress dialogue more aggressively than general-audience subtitlers to make room for the extra information.
The one-frame problem
In professional broadcast subtitling, cues are timed to the frame. At 24 fps, one frame is 41.67ms. At 25 fps, 40ms. At 23.976, 41.71ms.
A cue that starts one frame early feels slightly rushed. One frame late and the first syllable of dialogue is missed. In practice, most broadcast guidelines allow 1–2 frames of tolerance, but professional QC can reject a file that is off by more than that.
For web video, one-frame precision matters less. HTML5 subtitle rendering has its own lag, and viewers don't perceive differences of 40ms. But for cinema and broadcast delivery, one frame is real.
Time vs. feel
The most important timing lesson: a technically correct cue can still feel wrong.
A cue that starts on time but ends too abruptly. A cue that is perfectly readable but has no breathing room before the next one. A cue that is dense but comes with a slight pause afterwards — inviting the viewer to exhale. All of these are timing decisions that no guideline covers.
The best subtitlers are not the ones who know the guidelines best. They are the ones who feel the rhythm of a scene and adjust accordingly. They watch the same three seconds of video repeatedly until the cue feels right. They sometimes break the rules — a cue slightly longer than the guideline, a lead-in slightly larger than standard — because the scene asks for it.
The invisible work
The audience never notices a well-timed subtitle. They just read and understand. The timing is doing its job by being invisible. That invisibility is the goal. Every tool that helps subtitlers reach it faster — including the ones on this site — is a tool that respects the craft.