You set the duration you want and Aistube writes your faceless video script adjusted to that word count, divided into scenes with hook, development and closing.
A script that "more or less" lasts one minute doesn't work: you either have thirty seconds of voice-over left over or the video falls short and flat.
In short: you tell Aistube how long you want the video to last —between 15 seconds and 20 minutes— and the script comes out with the number of words that duration requires. The calculation is explicit: 150 words per minute of voice-over, with a ±10% margin that the AI is instructed to respect. For a 60-second video, the budget is 150 words; for a 30-second Short, 75. It's not a post-hoc estimate: the budget enters the prompt before writing the first sentence.
The script doesn't arrive as a text block: it arrives already split into scenes. Aistube calculates how many are needed at a rate of about 8 seconds per scene —the typical duration of a generated video clip— with a minimum of 3 and a cap of 30, and you can set another number from the advanced options within that same range. Each scene brings its narration, what's seen on screen, in which location it takes place, which characters appear and a camera movement prompt in English to animate it later, all in the same generation.
The structure is three acts and is marked scene by scene: act 1 hook and setup —the AI is instructed to win the video in the first three seconds—, act 2 development with twists and act 3 climax, closing and soft CTA. If your channel has a visual bible, the exact names of your characters and locations are injected into the prompt with their anchor description, so the script talks about your real cast and not about a generic "a man". And if you've chosen a format template —top 5, curiosity, true crime, tutorial, conspiracy, storytelling, finance or psychology— that format's structure also goes into the instructions, along with the channel's niche and tone and, if the project was born from a saved idea, its hook, summary and angle.
When finished, Aistube counts the actual words in the script, estimates the voice-over duration you'll get and warns you in the header if you've deviated more than 15% from the target, so you can correct it before spending a single token on images. You can add your own free instructions —up to 1,000 characters— to direct the tone, focus or whatever you want to prohibit. Aistube is the mini-app from AIsToHire to produce faceless YouTube channels from start to finish, and this functionality is used at https://aistohire.com/aistube-faceless-youtube-channels-ai-121/f/ai-script-generator-faceless-videos
Explicit word budget
150 words per minute with ±10% margin. The target duration converts into a concrete number of words before writing, not after.
Automatic scene division
one scene every ~8 seconds, between 3 and 30, the size that best fits a generated video clip. You can set the number yourself from advanced options.
Three acts marked scene by scene
hook in the first 3 seconds, development with twists, and closing with payoff and soft CTA. Each scene knows which act it belongs to.
Your real cast, with their names
the characters and locations from the channel bible enter the prompt with their anchor description, so the script names them as is and the scenes recognize them.
Eight format templates
top 5, curiosity, true crime, tutorial, conspiracy, storytelling, finance, and psychology. The format structure adds to your channel's niche and tone.
Free instructions up to 1,000 characters
direct the focus, the register, or what you don't want to appear, without having to rewrite the script by hand.
Deviation warning
Aistube counts the words, estimates the voice-over duration, and marks the header in amber if you drift more than 15% from the target.
One generation, complete package
each scene comes with voice-over, on-screen action, location, characters, and camera movement prompt in English, ready for storyboard and images.

Create your Aistube account, choose your video duration, and let AI deliver it in three acts divided into scenes, ready for storyboard.