Voices and narration

Every Vidext module speaks. Narration is generated with the module — written scene by scene as part of the build, then synthesized in the voice you choose.

Choosing the voice and presenter

Voice and presenter avatar are chosen together in the AI instructor picker: pick the avatar that delivers the module and the voice it narrates in, and preview a voice before you commit to it. Your selection applies when the module is built.

Your organization has a voice library:

• Premade voices — ready to use, professionally neutral.
• Cloned voices — your own. Upload clean voice samples and Vidext creates a custom voice for narration. A cloned voice goes through a short training step before it is ready.

Voice cloning is how a course can sound like your Head of Sales without booking your Head of Sales for a recording day.

How narration works

• The build writes a spoken script per scene — written for the ear, not copied from the slides — then polishes it for natural delivery and synthesizes it.
• Narration drives the learner experience: pacing, auto-advance, and subtitles (generated automatically, off by default, toggleable by the learner).
• Learners control playback: pause, navigate, and adjust speed from 0.5x to 3x.

From narration to presenter video

Scenes using presenter layouts can show a talking avatar delivering the narration. The presenter render starts only after you choose Publish, using the exact narration you reviewed in the draft. The candidate version does not become live until the render and final verification succeed. See Build, review, and publish.

Editing what is said

Narration is content like any other: ask for changes by chat ("make the intro narration shorter and warmer") and the audio is regenerated for the scenes you touched. See Editing published content.