# Voice and avatars

> Script, voice, pronunciation and on-screen presenter.

# Voice and avatars

Narration is written and voiced separately from the recording. That separation is what makes a
correction cheap — you change a line, not a take.

## The script

Each scene carries one narration line. Edit them in the script view, which reads top to bottom like
a document.

**Rewrite with AI** rephrases a selected line — for tone, length or clarity — without touching the
rest. Useful for tightening a line that runs long against its scene.

## Choosing a voice

Pick from the voice catalogue, or keep the original audio from your recording if your own narration
is already right. A workspace can set a default voice so every new project starts on-brand.

Listen to one scene before generating them all.

## Pacing and pauses

Pacing is a project-level setting. Insert pauses where the viewer needs a beat — after a result
appears, before a new section.

## Pronunciations

Product names and acronyms are where synthetic voices fail. Add them to the pronunciation list once
and every project in the workspace gets it right. Don't spell words phonetically in the script —
that fixes one line and corrupts your captions.

## Avatars

An optional on-screen presenter. Choose an avatar from the catalogue and set its placement in the
frame.

Avatars are the most expensive thing in a render, so they are worth previewing on one scene before
applying to a whole project. Cost shows up in the estimate before you approve a job.