Instant voice cloning

Definition

Instant voice cloning, also called zero-shot cloning, creates a reusable voice from a short sample. You provide 10-30 seconds of clean speech and a verified consent recording, then use the returned voice ID to synthesize with a supported model. This is the self-serve path on paid plans.

The sample

Quality starts with the sample. Aim for 10-30 seconds of clean speech, under 60 seconds and under 5MB, recorded without background noise. This voice sample is separate from the required spoken-consent recording.

Self serve

Instant cloning runs through the API or the Console without a sales conversation. Send the sample and consent, receive the voice ID, and synthesize. That makes it the right fit for developers who want to add cloning to a product and for users who want their own voice quickly.

Consent

Create a consent challenge and record the same speaker reading the returned phrase. Send the challenge ID and recording with the voice sample. Instant does not mean skipping consent; it means a short sample and no fine-tuning step. See the consent guide.

When to go professional

Professional cloning fine-tunes on hours of a speaker’s audio and is arranged with our team for signature narrators and brand voices. For self-serve cloning without a fine-tuning run, start with the voice cloning API.