[Feature request] Possibility of using a fixed seed for inference in XTTS V2

[Mixomo]

🚀 Feature Description

Possibility of using a fixed seed for inference in XTTS V2, so that the consistency and robustness of the model could be better evaluated with different reference audios and/or modification of parameters.

Thanks!

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.