How many words fit in a generated shot
A shot with a fixed ceiling can hold a predictable number of spoken words. Published measurements of conversational speech and published subtitle limits give the two ends of the calculation, and generated shot lengths give the third. As of 2026-09-12.
| Figure | Value | Where it comes from |
|---|---|---|
| Conversational speech rate | About 196 words a minute overall | Switchboard telephone corpus |
| Sora 2 clip lengths | 16 or 20 s, extendable to 120 | OpenAI video generation guide |
| Subtitle reading limit, adult | Up to 20 characters a second | Netflix partner help |
| Subtitle reading limit, child | 17 characters a second | Netflix partner help |
| Veo 3.1 clip lengths | 4, 6 or 8 s | Gemini API |
Inclusion rule. Published measurements and published limits. Rates quoted without a source, or derived from a single observation, are not listed. Order. Alphabetical by figure.
1The arithmetic is short and nobody does it
At roughly three words a second, a six-second shot holds about eighteen words including the pause before and after. Most dialogue written without that constraint runs longer, and the result is either a rushed read or a line that does not fit the shot it was written for.
Doing the sum while writing is far cheaper than discovering it at generation, because the fix is a rewrite either way and at generation it also costs a call.
2Subtitles have their own ceiling
A line that fits the shot can still be unreadable as a subtitle. The published character-per-second limits are lower for younger audiences, which matters for a format distributed by app where the audience skews wide.
3Pauses are part of the count
Measured speech rates including silences run well below the rate excluding them. A shot list that budgets words at the faster figure produces lines with no room to breathe, which reads as hurried even when every word fits.
4Where the numbers come from matters
Speech rates measured on telephone conversations are not the same as scripted dramatic delivery, which runs slower. Using a conversational figure to size dramatic lines under-counts the time needed, usually by enough to matter in a six-second shot.
The published figures are a starting point to be adjusted against a production's own delivered shots, not a constant to be applied.
5Subtitle limits bind a different audience
A line that plays comfortably can still exceed a reading limit once subtitled, and short-drama apps are watched muted often enough that the subtitle is the primary channel. Sizing to the shot and sizing to the subtitle are two constraints, and the tighter one wins.
6Sources read for this entry
This page defines the term and logs what each vendor documents about it, each figure read on 2026-09-12. Related: Lip-sync, Voice id.