Create music with your own verified voice

The generator’s Upload your voice entry is for a voice you own or are fully authorized to use. It is not a tool for imitating another person or implying an endorsement.

The feature appears with supported models and eligible paid plans. A fresh upload and verification may be required for each song creation.

Prepare the source sample

Use a clean clip with at least 10 seconds of singing or speaking. A dry recording is usually easier to verify than audio with loud music, reverb, crowd noise, or several speakers.

Before continuing, the consent step asks you to confirm:

• the recording is your own or you hold all necessary rights;
• you will not impersonate someone without clear permission;
• you understand how voice-related data is processed under the Terms and Privacy Policy.

Choose Upload audio or Record audio, then check that the clip contains a steady voice level and no long silent opening.

Complete phrase verification

The studio displays a short phrase. Choose Record phrase, read the complete sentence exactly, and submit it before the countdown expires.

If verification fails:

• match the displayed words rather than paraphrasing;
• speak clearly at a natural pace;
• use the same voice as the source sample;
• move to a quieter room or upload a cleaner file;
• restart if the verification timer has expired.

When the status becomes Voice ready, return to the generator and finish the song settings. A verified voice is only one input: Style, lyrics, vocal range, and arrangement still affect the performance.

Keep expectations and rights clear

Do not describe the result as a literal recording of a person who did not perform it. Review pronunciation and any unintended identity cues before publishing. Delete a saved voice from the voice library when you no longer want it available; songs already created with that voice are not removed automatically.

If you need a reusable musical identity derived from a finished song instead of a newly verified source sample, use Music Persona. For the first test song, follow the basic generation workflow.

Before publishing, document the authorized voice source, the consent basis, the final review date, and who approved release. Confirm the lyrics, accompaniment, artwork, and other assets have their own permissions; authorization for a voice does not cover unrelated material. Keep the test and final versions clearly named so an earlier unstable take is not distributed accidentally.