All tutorials

Pandrator · tutorial

Use more than one speaker or voice

Assign a narrator and character voices, or map speakers from a recording to a voice cast.

Updated 13 October 2026 · Pandrator 0.12.0

Screenshots show the running app. Some come from the development build; newer controls are noted at the relevant step.

  1. Choose Multiple voices

    In the project overview, find Voices and select Multiple voices. One narrator uses the project voice throughout; switching modes does not require deleting your cast.

    Narrator and Fred cast settings in a Christmas Carol project
    A narrator and character roles have their own references and assignments.
  2. Identify the speakers

    Use existing source-speaker attribution when available. For a book, review dialogue attribution or assign it manually. Keep the narrator distinct from named characters and check a passage where the speaker changes.

  3. Assign the cast

    Bind each speaker or character to a voice. A compatible managed voice can be reused from the voice library. Check every role: an unassigned character can fall back to the narrator.

    Reusable synthetic voices in the app voice library
    Saved references can be reused in compatible projects and models.
  4. Generate a short conversation

    Generate a few consecutive lines and listen for speaker changes. Pandrator assembles separate requests for the cast; this does not require a model with native multi-speaker input. Adjust the cast, pauses or selected takes before the full run.

    Generated narration with its speech text and completed take
    Check the speech text and the selected take before assembly.

Check the result

  • Check who speaks each line, not just whether the voices sound different.
  • Make sure the voices use a model and language they support.
  • Keep voice assignment, pronunciation and acting directions as separate decisions.