Sixty seconds, and you don't set it. Duration is derived from script length and natural speaking cadence, with a live word count and duration estimate updating as you type. Six words reads as roughly three seconds, so sixty seconds is somewhere near 120 words. There's no slider and no setting that compresses a long script into a short runtime — the only way to change the length is to edit the script.
Most generation tools give you a duration control and let you fill it. This one works the other way round, which changes how you write rather than just what you set.
Why is there no duration setting?
Because a person speaking takes as long as the words take.
A duration slider makes sense when you're generating motion — the model can fill four seconds or thirty with whatever the prompt describes. It doesn't make sense for speech, because the words have a fixed length and speeding them up reads as rushed rather than shorter.
So Talking Actors derives duration from the script instead. You write, and the runtime follows.
The practical consequence: there is no setting that will compress a long script into a short clip. If a script would run ninety seconds, the answer is to cut it rather than to configure your way out.
The script is the runtime control. Nothing else changes the length.
What does the live estimate give you?
A word count and a duration readout that update as you type.
That's more useful than it sounds, because it turns writing-to-length from a guess into something you can see. If a clip has to land at fifteen seconds, you write until the readout says fifteen rather than writing, generating, and discovering.
The calibration is roughly six words to three seconds. Which gives you a rule of thumb worth internalising:
- 6 words
- ~3 seconds
- 12 words
- ~6 seconds — one line of short drama dialogue
- 30 words
- ~15 seconds
- 60 words
- ~30 seconds
- 120 words
- ~60 seconds, at the cap
Where this falls short. Those are derived from the stated six-words-to-three-seconds ratio rather than measured across scripts. Delivery pace varies with punctuation, sentence length and emotional register, so treat them as a planning guide and trust the live readout over the table.
What happens at sixty seconds?
It's a hard cap, and the pacing feedback operates up to it.
For vertical drama that's rarely the binding constraint. Episodes run thirty to ninety seconds, and a talking-head beat inside one is usually a fraction of that — the piece ends well before the ceiling.
Where it binds is anything presentation-shaped. A product explanation, a founder update, a tutorial intro. Those overrun sixty seconds easily, and at that point the format is the wrong one rather than the length being the problem.
How do you write to a length?
Four moves, cheapest first. These are the same cuts that shorten a line, applied to a whole script.
Cut the address and the wind-up
Greetings, names, "so today I wanted to talk about." A talking-head clip has no runway — the first words are the first thing a viewer hears and they should carry information.
Cut the hedges
"I think," "maybe," "kind of." At this length hedging reads as padding rather than as character, and it costs you seconds you need elsewhere.
Split into two clips
Two thirty-second pieces frequently outperform one sixty-second piece, and they're easier to generate and easier to re-cut. This is the move people reach for last and should reach for earlier.
Move information to captions or visuals
Anything the viewer could read or see doesn't need saying. Most vertical viewers watch muted anyway, which makes this the cheapest cut available.
Does script length affect cost?
Yes, and it's the main driver.
Credits scale with script length, so the same editing that shortens the runtime also lowers the price. One action moves both, which is unusual — in most tools length and cost are set separately.
The second driver is the actor. Casting a real human creator adds a small flat fee; AI actors and an actor you made from your own photo don't. Multi-look creators sit at a higher shared rate than single-look ones, and all multi-look creators share that same rate — so a five-look creator costs the same as a two-look one.
The cost shows before you generate, so none of this has to be estimated.
What about pacing within the clip?
Two controls, and one of them is more precise than a prompt would be.
Emotion control has two modes. Auto infers context from the script and renders matching expression and vocal inflection across the whole thing, with no tagging. Manual takes inline triggers inserted at specific words — a forward slash at any point in the script editor.
For a clip written to a tight length, Manual is the useful one. If a beat has to land — the pause before an answer, the shift on one word — you can place it exactly rather than adding words to signal it. That buys back seconds.
You can also audition the voice before generating, via the speaker icon on each creator card. Worth doing when the read matters, because voice affects perceived pace as much as word count does.
- Sixty seconds is the cap, and you don't set the duration. The script determines it
- No setting compresses a long script into a short runtime. Cutting is the only lever
- Roughly six words to three seconds. So sixty seconds is somewhere near 120 words
- Trust the live readout over any word-count table. Delivery pace varies with punctuation and register
- For vertical drama the cap rarely binds. For presentation-shaped content it binds immediately
- Past a minute, use a different tool. Make a Custom Video runs to thirty seconds on a slider; Creative Studio goes further
- Credits scale with script length, so one edit moves both the runtime and the price
- Place a beat with an inline emotion trigger rather than adding words to signal it
Sixty seconds. Duration is derived from script length and natural speaking cadence rather than set directly, with a live word count and estimate updating as you type. Pacing feedback operates up to the cap.
No. There's no duration slider, and no setting compresses a long script into a short runtime. The script is the runtime control, so the only way to change the length is to edit the words.
Somewhere near 120, working from the stated ratio of roughly six words to three seconds. Treat that as a planning figure and trust the live readout, since delivery pace varies with punctuation, sentence length and emotional register.
Cut it, split it into two clips, or use a different tool. Make a Custom Video runs four to thirty seconds on a slider, and Creative Studio's per-model durations and Extend rounds go further. Two shorter pieces also frequently outperform one long one.
Yes. Credits scale with script length, so the same editing that shortens the runtime lowers the price. The other driver is the actor — a real human creator adds a small flat fee, and multi-look creators sit at a higher shared rate than single-look ones.
Yes, with Manual emotion control — inline triggers inserted at specific words using a forward slash in the script editor. That's more precise than describing intent in a prompt, and it lets you place a pause or a shift without adding words to signal it.
The script is the runtime
Write, watch the word count and duration estimate move together, and place a beat exactly with an inline trigger. Plan First previews the script without consuming credits, and you can audition the voice before you spend anything.
Try Talking ActorsMore on tools, features and how the platform fits together in Product.