Clone a voice
Create a copy of your voice, or a voice you have permission to use, from a short sample.
Updated November 10, 2026 · 1 min read
On this page
Before you start
You'll need 10 to 15 seconds of clear speech from one person. For the best match, record somewhere quiet, with no music or background noise.
You may only clone your own voice or a voice you have permission to clone, and it can't be a public figure.
Create the voice
- Open Voice Cloning and click Clone a voice (or New voice).
- Upload a file or record a sample. Recording stops by itself at 15 seconds.
- Give your voice a name, up to 60 characters.
- Choose the language it speaks.
- If you uploaded a file, confirm the two statements about permission and our terms.
- Click Save voice.
Saving a voice doesn't use credits, but it does use one of your voice slots.
Sample requirements
- Formats: MP3, WAV, M4A, WebM, OGG or FLAC.
- Size: Up to 10 MB.
- Length: 10 to 15 seconds works best. Shorter samples give a less close match. For longer ones, we use about the first 15 seconds.
Generate with your voice
Select your voice, write your script and click Generate speech. Your settings work the same way as in Text to Speech. The credit rate depends on the model, and the button shows the cost before you start.
Voice slots
Cloned voices, blends and designed voices share your plan's slots. Before you save, the page shows how many you've used. If they're all in use, delete a clone in the Voice Library to free one, or compare plans to get more.
If your voice is in review
Some new voices are reviewed first. You'll see "Voice saved. It's being reviewed", and the voice appears in your list once it's approved. A voice that goes against our content policy can be blocked and is then no longer available.