Bill Wurtz (Talking)

for those who are curious the dataset was 22 minutes long

no yes

maybe so

we need the model to do the entire history of the entire world, i guess video

I don't really have any other ideas for what to add as samples

this works good for monotone voices if you wanna use it

It's really only good for speech though I tried using various songs and it's pretty bad

oh

like I said in the thread description

srry(i thought it would work bc my royalpear model works with singing but was trained on 15:30 of talking)

Don't worry about it :)

Here's another talking example (Audio is from Sam O'Nella)

The sun is a deadly laser!