Natalya Hart (WWE Wrestler)

Example: https://drive.google.com/file/d/1xEabJR8CJlL8yAMMuVmLlRLM48U1nNnv/view?usp=sharing

This is indistinguishable from her real voice. How do your models sound so good on weights? Many of the models created here are low quality, probably because of the lack of epochs. But yours always sound great! What's your secret, if you don't mind me asking?

I think Weights has a way of automatically stopping the epochs just when it hits the right amount. That's why it doesn't take long to train these models. But I also de-reverb my models. If you leave reverb, effects or background vocals in your dataset, it will become part of the training process and end up in your model. You need the cleanest acapellas you can find with absolutely zero effects or interference.

Oh okay. That makes sense! Thank you. I'm about make those attempts.