Two starting points
Quick Clone works from a short recording (5-10 seconds) and is the fast way to try a voice before it becomes a series. Studio Clone accepts a longer source (10 seconds to 5 minutes) and is built for narration that needs more range across chapters or episodes.
Both paths produce a private voice that only appears in your own Voice Picker once you accept its preview - it is never shared across accounts.
- Record in a quiet room with one identifiable speaker.
- Avoid music, overlapping voices, or long silence in the source file.
- Confirm consent before any recording is processed - this is a required step, not a formality.
Preview before it enters your library
Processing produces a preview you listen to before accepting the voice into your library. Nothing is added to Voice Picker until you accept it, so a source recording that didn't turn out well never leaves an unusable voice behind.
Using it in a long-form take
Once accepted, a cloned voice behaves like any catalog voice: select it, generate every section with the same settings, and use section regeneration the same way you would with any other voice if a passage needs to change later.