Confirm this is a new feature request

• I've checked the current issues, and there's no record of this feature request

Describe the feature

Creators should be able to provide their sample voice either by uploading directly or by extracting samples from their existing videos. They should then be able to use models like:

1. Elevelabs
2. xtts-v2
3. f5-tts

For audio generation, creators should simply provide a script along with the model/engine name, and the system should generate audio in the cloned voice.
This feature can integrate seamlessly with video generation and editing agents to automate the entire video creation pipeline.

Note: Think about the copyright issues if user is using unauthorized voice.

Additional Context

No response