Skip to main content
Your voice clone lets you generate AI voiceovers that sound exactly like you. It’s useful for creating content without recording every time, narrating footage in post, or dubbing your videos into other languages in your own voice.

​
Setting up your voice clone

Your voice clone is created as part of the AI Twin setup process. If you’ve already set up your Twin, your voice clone is ready to use.
1

Record your calibration clip

If you haven’t set up your AI Twin yet, go through the AI Twin setup guide. During setup, you’ll record a 1-minute calibration script. This is what Captions uses to clone your voice.
Record in a quiet room, close to your microphone or phone. Avoid reverb, echo, and background noise. The quality of this recording directly determines how natural your clone sounds.
2

Speak naturally during the calibration

Read the calibration script at your normal speaking pace. Don’t over-enunciate or speak more slowly than you normally would. The clone learns your natural cadence, tone, and inflection, and speaking artificially will make it sound stiff.

​
Using your voice clone in a project

1

Open Voiceover in the editor

In the Captions editor, tap Voiceover from the toolbar or the audio panel.
2

Select AI Voice, then My Voice

Choose AI Voice from the voiceover options. Your voice clone appears as My Voice once it’s set up. Select it.
If My Voice doesn’t appear, your AI Twin setup may not be complete. Return to the AI Twin setup flow to finish the voice clone step.
3

Type or paste your script

Enter the text you want spoken. This can be a full narration, a short CTA, or any portion of your video’s audio that you want to generate rather than record live.
4

Generate the voiceover

Tap Generate. Captions produces an audio track in your voice. This typically takes a few seconds for short scripts, slightly longer for full-length narrations.
5

Review and adjust timing

The voiceover track is added to the timeline. Listen through the full clip. If the pacing feels off, you can trim or reposition the audio track, or regenerate the voiceover with slight edits to the script (adding or removing words can adjust the pacing).
6

Layer it with your footage

Your voiceover sits on the audio track in the timeline. You can layer it under footage, B-roll overlays, or use it as the sole audio in a Prompt to Video project.
For Prompt to Video specifically, a voiceover recorded in your voice clone can drive the avatar’s lip sync, producing a complete video that sounds and looks like you without any filming.

​
Voice clone for dubbing

Your voice clone can be used when dubbing your videos into other languages. Instead of a generic AI voice, the dubbed audio will sound like you, preserving your vocal identity across languages.
1

Enable voice clone in Dubbing settings

When dubbing a video, open Dubbing, select your target language, and look for the voice selection option. Choose My Voice to use your clone for the dubbed audio track.
2

Review the dubbed output

Your voice clone speaks in the target language using your natural tone and cadence. Review the dubbed clip and check that the pacing sounds natural. AI dubbing with a voice clone is very accurate but subtle corrections may be needed for some phrases.

​
Tips

  • Record your calibration clip in the same environment and at the same mic distance each time you recalibrate
  • Speak at your natural pace. The clone reproduces your cadence, so if you record slowly it will sound slow
  • The clone handles emotion and expressiveness based on your source recording. If you sound flat during calibration, the clone will sound flat in output
  • For best results, re-record your calibration if you’ve changed your speaking style or recording environment significantly

​
What’s next?

AI Twin Setup

Set up your AI Twin and voice clone from scratch.

AI Voiceover docs

Full reference for the AI Voiceover feature.

No-Camera Video

Use Prompt to Video to make talking-head videos without filming.

Dub a Video

Dub your video into another language using your voice clone.
Last modified on April 20, 2026