Record half a minute of yourself, and every voiceover in your workspace can sound like you: a single line on a clip, or a forty-segment script read end to end. No booth, and no re-records when the copy changes.
Set a voice up once for the workspace
A voice belongs to your workspace rather than to one project. You make it once, from a live recording or an audio file you already have, and it waits on every surface that can speak.
A passage is written for you, so you never have to invent something to say
Or upload a recording you have: an old voiceover, an interview, anything with clean speech
A clip with no clear speech in it is refused up front rather than cloned into nonsense
Change the script later and the read is one click, not another session at the mic
Draft in the cheap voice, deliver in the good one
Three models sit behind one voice. They differ in how convincing they sound, what they cost, and what happens to the voice after you make it.
One works the instant your recording exists, with nothing stored anywhere
One hands us a voice file we keep, so deleting it is ours to honour
The best-sounding one is held by the provider and lapses if unused. We keep it alive for you, and your recording can always rebuild it
Set one voice up on all three and switch per generation
A full script, read line by line
The payoff is the forty-segment explainer you would otherwise narrate in one sitting and re-narrate every time a sentence changed. Write the script, pick your voice, and take the lines one at a time.
Every segment gets its own take, so fixing one sentence re-reads one sentence
Each take lands on an audio lane at the segment it belongs to, already in position
Mix it like any other clip: duck the music, set the levels, keyframe the volume
Or use it for a single line on a single clip, from the same picker
Questions
Whose voice am I allowed to clone?
Your own, or someone else’s with their explicit written permission. You choose which when you set the voice up, and that choice is recorded with the wording you agreed to. Cloning a voice without permission is unlawful in many places and a criminal offence in some, and the account that made the voice is responsible for how it is used.
How much of my voice do you need?
About thirty seconds of clear speech. A passage is provided so you do not have to think of something to say, and anything under twenty seconds is refused because every model clones badly from less.
Do you check that the voice is mine?
We can confirm the recording was made live rather than downloaded: we show a phrase, you read it aloud, and we check it was spoken. That blocks a voice lifted from somewhere else, but it is not identity verification, and we do not claim otherwise.
Can I delete a cloned voice?
Yes, and it removes your recording, the voice files we hold, and every model set up from it. One model keeps the voice on the provider’s side, where it self-deletes after a week without use; deleting the voice here stops us keeping it alive.
Which voice model should I use?
The recommended one for anything other people will hear: it was the best of four we tested, and the fastest. Use the cheap one for drafts and temp tracks, where per-word cost matters more than the last few percent of realism.
Can I use a cloned voice for a full script?
Yes, and that is the case it was built for. Write or paste a script, pick your voice once, and read each timed segment with its own take that lands on the timeline in position.