i generally call it "list.txt"

ohh ok it just clicked what you said bout "wavs/filename.wav|This is where the transcript goes." thanks i get it now

👁️👄👁️

hello again i got a miss shape error i downsampled to 22050 with pydub AudioSegment then just re saved them to the wav/ folder idk if thats why the shape is how it is but should i just flatten it to one dim? RuntimeError: shape '[1, 1, 611535]' is invalid for input of size 1223070

or its not mono sorry this is my first time working with audio data?

Make your audio files all mono

and 22KHz

hello @User thanks i do a lot of nlp work with gpt2 to generate text from everything from cartoons to rappers. Im actually going to make my voice and add the text gen to so its a full ai Rick sanchez I was wondering if you have a github page for the project i love fun ai projects

That sounds great! But no, currently uberduck doesn't have a github repository.

word where are you hosting all these models from just wondering have been trying to deploy some of my models as server less function just for inference but too many problems with cold starts and general speed

AWS

yeah just some ec2?

unknown

Zwf runs the site

yes best way to do it as a software dev imo lol

lol

Does anyone know on what the melee announcer is saying here?

Fighting Wire Frames!

Huh?

Alright

XD

Ngl this is just weird

and I couldn't understand what he said

source

ai

it works wonderfully

what does Kernel_Size Do?

determine the convolution size of the kernel in the postnet of tacotron

yeah are the notebooks on github or anything? would love to add more hyper parmas to tune like epochs and lr for starters this is sick tho

Check the official Uberduck notebooks for the links to the original notebooks, which are open-source.