Michael Jackson (BAD Era) *IMPROVED MODEL*

Sounds good! What songs did you use?

Thanks for sharing. I've made some BAD models but none turned out this good.

Another reason I asked about the dataset was because I don't think the weights model do well. The ones where people create the models themselves and upload them here sound so much better.

weights voice training is low in quality, but I don't think a lot of people who uses it knows how to de-reverb anything.

Thanks for answering! Once last question for you. I noticed that weights doesn't use many epochs. Is that part of why the training is low quality?

It doesn't train anything to a high enough number of epochs, it used to show the epochs at first, it would be anywhere between 50-100, which is not enough imo.