Skip to main content

Voices

Voices are used for any Gabber feature that can synthesize speech by passing in the voice id where appropriate.

Gabber ships with large list of default voices. Gabber also provides voice cloning for creating voices.

Testing voices

The Voice tab can be used in the Gabber dashboard to test how a voice sounds. Voices can also be tested in dashboard playgrounds.

Creating a Zero-shot Clone

You can upload a short (15s) speech mp3 file to clone a voice. This will create a voice object only visible to your project.

Voice cloning is available in the dashboard and the cloneVoice API.

Creating a Premium Clone

We also support premium clones. The clones require more data (~30min of clean audio) and they sound much better. Premium clones are dedicated fine-tunes for your project.

Because they are fine-tunes, they can also be used to create custom emotion tags like (). Premium clone inference adds no extra cost. Currently, these clones are not self-serve, so reach out to us on Discord to get started.