Casting Multiple Voices
A multi-voice audiobook is what separates a narration from a performance: the narrator in one voice, each character in another. In VocalLab this is handled per line, which makes dialogue straightforward and keeps you in control of every delivery.
New to the audiobook editor? Start with How to Make an Audiobook — it has a full video walkthrough.
How casting works
There is no separate "character list" to set up first. Every speaker block carries its own voice. You write a line, you pick who says it, and you move on.
To cast a line:
- Click the voice chip at the top of the speaker block
- Pick from the strip of voices already used in this project, or open the full picker for anything else
- That's it — the line is cast
Picking a voice on any line automatically adds it to the project's voice list, so the second time that character speaks they're one click away.
A line with no voice yet shows a Pick a voice warning chip. Those lines won't generate, and they'll block a chapter generation until you cast them.
Where your voices come from
Any voice you can use elsewhere in VocalLab works here:
- Built-in voices — the library
- Cloned voices — your own voice, or one you have rights to
- Designed voices — invented from a written description
Voice design is particularly well suited to audiobooks: you can spin up a distinct voice per character from a sentence of description, without needing a recording of anyone.
Managing the cast
Open the Settings panel and find the Voices section. It lists every voice in the project and gives you two operations:
Change — swaps that voice across the entire project in one action. This is the one to reach for when you've narrated three chapters and decide the narrator should sound older. It is not limited to the current chapter.
Delete — removes the voice from every line it was cast on, but keeps all the text. Nothing you wrote is lost. Those lines simply become uncast and need a new voice before they'll generate.
There's also Add another voice to bring a voice into the project before you've used it anywhere.
Same voice, many lines
Reusing one voice across dozens of lines is normal and expected — that's how a narrator works. There's no penalty and no limit on how many lines share a voice.
Emotion and delivery
Casting sets who speaks. Two more controls set how.
Emotion — each speaker block has a dropdown: happy, sad, angry, surprised, fearful, disgusted, laughing, or whispering. Set it per line, so a character can be cheerful in one breath and furious in the next.
Vocalizations — drop these inline in the text where you want them: [laugh], [breathe], [clear_throat], [cough], [sigh], [yawn]. The vocalization bar appears when a block is focused. Used sparingly they do more for realism than almost anything else.
Tips
- Cast the narrator first and generate one test line. Committing a whole chapter to a voice you haven't heard in context is the most common way to waste points.
- Keep character voices clearly distinct. Two similar voices in a back-and-forth are hard for a listener to follow. Differences in pace and pitch read more clearly than differences in accent alone.
- Use pause blocks between speakers in fast dialogue. A half-second gap does a lot for intelligibility.
- Changing a line's voice marks it stale. It gets an Edited pill and the next chapter generation will redo it — so swapping a character's voice mid-project costs points for those lines.
Troubleshooting
"N lines don't have a voice — pick one for each before generating." Some lines are uncast, usually because you deleted a voice from the project. Find the blocks showing Pick a voice and cast them.
I deleted a voice and lost my work You didn't — deleting a voice keeps the text on every line. Only the voice assignment is cleared. Re-cast those lines and generate.
Changing one line changed the whole project You used Change in the Settings panel, which is project-wide by design. To alter a single line, use the voice chip on that block instead.
The same character sounds slightly different between lines Each line generates independently, so there's some natural variation. Keeping the same voice, model, and emotion for a character minimizes it.
Related
- How to Make an Audiobook — the full walkthrough
- Exporting Your Audiobook — turning it into a finished file
- How to Design a Voice from a Prompt — invent a character voice