When cloning, how can we make the generated multiple audios consistent?

[dnlzsy]

When cloning, how can we make the generated multiple audios consistent?

[Respaired]

torch.manual_seed(some integer) right before running the inference function

[m-aliabbas]

use longform inference