- Run a sample from your approved speaker through your chosen tool using a real script.
- Clone ScriptIf you need to clone your voice to add new narration using an approved speaker, the voice cloning feature will be useful.
- Field Illustrative entry Original question Can I turn a product photo into a short video without changing the label?
- Ensure you have permission from the speaker to create the clone.
Select a voice cloning tool. Within your video create a short sample using the approved voice. Consider intelligibility, pacing, consistency between different texts, and quality of export. Ensure that you have permission from the speaker, and that it can be used in the commercial way intended within the terms and conditions of the provider. Consider what disclaimers are required by the distribution platform.
Considerations
When to Use Voice Cloning, Stock Voices, or Dubbing?
Include a Voice Clone for a consistent voice. For clear narration in an explainer video, a stock voice may be suitable. To translate an existing video, start with dubbing. Ask a fluent reviewer to check both the translation and recording.
What Makes a Good Voice Sample for Cloning?
Upload voice cloining tone or record a reference audio. Uploads must be at least 30 seconds long. It is recommended that uploads are three minutes long, with clear audio. They must be in mp3 or m4a format. Files must not exceed 25MB. One approved sample of one speakers voice with little background noise should be uploaded. If a calm, informative voice is required, ensure the reference audio is recorded with this in mind.

Which Tool Features Matter for Video?
| Buyer question | What to test before paying |
|---|---|
| Can I fix one line? | Change a price or product name mid-script. Check whether you can replace that line and how well it matches the surrounding audio. |
| Can I control pronunciation? | Test the product name, an acronym, and a number. See whether you can correct each one and reproduce the fix in a later take. |
| Does it fit my editing workflow? | Export a sample, import it into your editor, and listen with the actual music and effects. Check the available audio formats on your plan. |
| What will revisions cost? | Check whether failed takes, regenerated lines, and language versions consume credits. Compare the cost of an approved clip after revisions. |
| Who controls the cloned voice? | Check who can generate with it, how long the provider keeps reference files, and whether you can delete the sample and the saved voice. |
For localization, run one short segment through AI video dubbing in the language you actually plan to publish. Ask a fluent reviewer to check product names, translated claims, and whether the speech lands on the right visual cue. If the speaker is on camera, check mouth movement in the exported video.
20 Second Test
Create a short edit containing a product name, number, and a line which must be timed to a cut. Make a first version. Make a change to the text and replace in the same section.
Example test script
Introducing the Luma Mini. Tap the base one time for a soft light. This starter kit is $39 this week. See what comes in the box.
Create a second version of this text with 49 instead of 39. Make sure this text still fits with the same clip and sounds similar. Play the full version on headphones and on a phone. If the product name is clear, the instructions match the text on screen and the edit flows well, mark as ready. Otherwise, if basic changes to pronunciation and timing are required, create a version of this clip using a stock voice, or a new human voice. Choose an option which means you are able to create an approved version within your time and budget.

Approval for Commercial Use
Speaker approval
Ensure you have written approval to create the voice clone, and to use the output. Include what campaign, platforms, languages, length and who can generate new text in the approval. For example, if you have approval for an organic product demo, you may not have approval to use in paid ads, translated versions or future campaigns. Ensure you also have permission to upload the reference voice.
Provider terms
Find out what terms apply to the platform usage, and your plan/voice model. On Xelta, the user is responsible for ensuring they have the right to use the inputs. If you are working for a client, ensure that you can provide the final video to them. Check if they can generate new content using the saved voice.
Distribution Requirements
Check what requirements apply to the final video based on the upload platform. On YouTube, if you are using a clone of your own voice for a voiceover/dub, no disclosure is required. However, if realistic AI content is used to make it appear that a real person said something they did not, a disclosure is required. Keep a note of the approved text, approvals and any disclosure made on the published video.

Conclusion
Ensure you have tested one approved voice in a real clip, edited one line and confirmed where the final video can be used. Save the approved version and approvals in your project. For further examples of AI video production, see AI video processes on LinkedIn.
Frequently asked questions




