How to make AI video with dialogue and sound

Use a video model that generates audio with the picture, write the dialogue into the script per scene, and keep scenes short enough for the lines to fit. Otherwise add a narrator voice over silent or illustrated scenes.

Two ways to get sound

Until recently most AI video was silent, and sound meant a separate text-to-speech voiceover laid on top. Some current video models generate dialogue, effects and ambience together with the picture, which makes characters feel like they are actually speaking.

Both approaches are useful. A narrator over illustrated scenes is cheap, clear and consistent. Generated dialogue is more alive but costs more and is less predictable, because the model performs the line rather than reading it exactly.

How AISP handles it

AISP’s Motion look uses a narrator voice over illustrated scenes with camera movement. Its video tiers (Draft, Cinema and Cinema Pro) generate their own dialogue and sound with the picture, which the app describes as “no robot narrator”. At the time of writing they cost 11, 31 and 53 credits per second; Cinema Pro renders full 1080p.

The script decides what is said. Each episode is written as scenes with dialogue before anything renders, and you can read it and ask for a rewrite, or write the episode yourself with your own lines.

The player can show subtitles, and the subtitle track can be downloaded as a .vtt file by long-pressing (or right-clicking) the subtitles button.

Tips for dialogue that works

• Keep lines short. A scene of 8 seconds fits a sentence or two, not a speech.
• One speaker per shot is more reliable than two people talking over each other.
• Put the emotion in the action line (“she whispers, furious”) rather than relying on punctuation.
• Use Draft to hear an episode before paying for Cinema or Cinema Pro.

Questions

Can I choose the voices?

On the video tiers the model performs the dialogue itself, so voices are not picked from a list. Motion uses AISP’s narrator voice.

Which languages can the dialogue be in?

Any of the 15 series languages, set when you create the series.

Product details checked October 2026. Rates and features can change; the app always shows the current cost before you confirm.