Sorry. It did work a little bit fine on the synthesis notebook, but didn't get added in on Uberduck.

Uberduck.ai really needs to stop putting voice models down. You think I'm right?

I hear you're doing AoStH Robotnik and DKCTV K. Rool soon. I can't wait to see them on the site!

That's why I'm gonna fix my Rocky Cocoa model since zwf had some troubles with the output

Im sure theres a good reason for it but I can understand how it would be frustrating

my chills model got in first try

I'm not meaning to put any voice models down, I completely know how hard this is. I just want the models that are public on the site to be able to basically synthesize full sentences.

And yeah I cant wait to finish them! I’ll start on Robotnik once I finish the others I wanna finish, though all K Rool needs is to have his audio clipped and then he’ll be ready for training.

the amount of time i spent trying to make chills is actually comical lmao

I want to add better model organization as well as a development dashboard that makes it easier to play with models while they're in development. I also need to work on some performance improvements that makes it more feasible to host a large number of models. Once I get this done I'm happy to host models no matter the condition.

Okay.

I understand.

Hey, uh, I tested the Ami voice and it almost kind of sounds off, and if you have it speak in full sentences, it speaks gibberish, so I'll just improve it soon. You have some thoughts about it?

I have a random question about the Half-Life scientist voice.

Have you guys ever considered or have used the podcasts that Harry S. Robins was featured in?

There's hours of him speaking, and his speaking voice is practically the same as his scientist voice.

Only problem is that the sound quality isn't exactly the best.

Dr. Hal has also read a bunch of poems in a cafe that has been recorded.

Never mind. I improved and submitted it now so it will be on the site soon.

Hi, question.

What exactly does the sigma in the synthesis tacotron thing mean?

And what happens if I increase or decrease the value?

you mean the sigma for the denoising?

I guess so, yes.

do you understand what sigma normally is (in statistics)?

No, I've never heard the term sigma before seeing it in the module.