How to add character voices to your AI audiobook

In Narratory, you can use one voice for the narrator and assign different voices to individual lines of dialogue. Start with a short scene, listen to the transitions, and refine the voices before generating the rest of your book.

By Asa Harland

1. Import your manuscript and review the chapters

Upload an EPUB or PDF, or enter text directly. Check the imported chapters and lines before generating audio. Unusual formatting can need manual correction; review who is speaking rather than assuming every line has been identified correctly.

2. Choose a narrator and a small cast

Choose the voice that carries the narration first. Then give your main characters voices that are easy to tell apart. You do not need a new voice for every person in the story: a smaller cast can make the listening experience easier to follow.

Pick a scene that contains narration and dialogue from two or three characters. It is a more useful test than an isolated sentence because you can hear how the voices work together.

Hear the difference for yourself

We’ve published multi-voice audio samples across fantasy, thriller, and nonfiction genres so you can judge the quality firsthand.

Listen to audio samples

3. Assign voices and generate a short sample

In the line editor, choose a voice for each dialogue line. Generate a sample and listen for clear changes between the narrator and characters. Adjust the wording, pauses or voice where needed.

Previews and revisions use generated-word credits. The free sample includes 500 generated words, so reserve some of that allowance for trying changes. Listening again to audio you have already generated costs nothing.

4. Review before generating the whole book

  • Can you tell who is speaking without looking at the text?
  • Are names and repeated terms pronounced as you intend?
  • Do pauses and voice changes make the scene easy to follow?
  • Does each character still sound suitable across different scenes?

If one line needs work, regenerate that line. Newly generated words use credits; the audio you keep does not need to be generated again. AI output still needs a listening pass before publication.

5. Export and check your distribution route

An active paid plan is required to create an export. Narratory includes ACX and Apple Books export formats, but technical formatting and retailer acceptance of AI narration are different questions. Confirm your distributor’s current rules for your title before committing to a full production.

Use only text and voice recordings you have the rights to use. If you clone a voice, make sure you have the speaker’s permission.

Compare the actual workflow, not a feature label

Character casting is also offered by other audiobook tools, including ElevenLabs Audiobooks. Compare the same scene in the tools you are considering: importing, assigning voices, fixing a line, reviewing the result and exporting. Check how each charges for previews and revisions.

Read our audiobook workflow checklist and production-cost examples to plan a complete book.

Hear your own writing before you commit.

Try 500 generated words with no credit card. Previews and revisions use your allowance; an active paid plan is required to create an export.

Try your 500-word sample