Voice-over
Voice-overs from a script, or a voice cloned from a reference video.
Every script on the canvas can be spoken. Pick from a library of AI voices, or clone one from a video you provide and reuse it across every video you make. The voice node sits between the script and the cut, so re-recording after an edit is one click, not a new project.
Two ways to get a voice
Cloned voices are per user. The number you can keep depends on your plan, from one on Starter to twenty on Studio.
- Pick a voice: a library of natural AI voices, in many languages, previewable inside the node
- Clone a voice: from a reference video or audio you provide; the clone is stored on your account and appears in your selector from then on
Built for long scripts
Long scripts are voiced in parallel chunks and stitched, so a several-minute voice-over does not take several minutes to come back. Transient provider errors are retried automatically instead of failing the whole recording.
Standalone text to speech
You do not need a video to use it. The text-to-speech tool takes any text, lets you pick or clone a voice, and returns the audio — useful for podcast intros, ads, or dubbing a script you wrote elsewhere.
Questions
- How much audio does cloning need?
- A reference video or audio clip with clean speech from one person. Longer, cleaner samples give better clones; background music in the sample hurts.
- Can I clone someone else's voice?
- Only clone voices you have the right to use — your own, or one whose owner has agreed. The terms of service cover this.
- Does the voice change when I edit the script?
- The voice node re-records from the edited script when you ask it to; the video shots downstream stay as they were unless you regenerate them too.