Personas and Voice References

Reuse the singer of one of your songs in new music with a Persona: choose a clear 10–30 second vocal section, name it, and pick it in the generator to keep the same voice across a project.

Last updated October 5, 2026

A Persona captures the singer of one of your finished songs so new songs can be sung in the same voice. It is the way to keep one artist's sound across an EP, a series of jingles or a set of extensions.

Create a Persona from a song

In the AI Music Generator, open Persona / Voice and choose Create reference → Persona.
Choose a song — one of your completed songs with vocals.
Mark the Vocal section: drag the slider or type a start and end so the window holds a clear vocal between 10 and 30 seconds. Use Play to check you selected singing, not an instrumental break.
Give it a Name, and optionally a Description (tone, range, texture, language) and a Style, so you can tell your Personas apart later.
Create it. It is usually ready in under a minute.

Pick a section where the voice is exposed and steady — a verse line over a sparse backing works better than a chorus buried in harmonies and effects.

Use a Persona

Open Persona / Voice in the generator and select it. New songs, extensions and covers are then sung in that voice. Remove persona clears the choice. A Persona has no effect on an instrumental, because an instrumental has no vocal to sing.

The style tags still decide the genre and arrangement; the Persona decides who is singing. To keep a series consistent, reuse the same Persona and keep the style lines close.

Your own voice

The Voice option — recording or uploading your own voice, verifying it, and singing with it — is shown only where it is available. Where it is not, create a Persona from one of your songs instead, or train a voice model for covers (see Voice Models and Covers).

Who can use it

Personas are private to your account. They are made from your own songs and appear only in your generator.