i really dont want to do that because when i go to the folder where the thing's supposed to be the folder is empty

Its saved on your google drive

i looked and its not there

wiat it is

omg okay i was being stupid

okay

okay i stopped it, thats terrifying

Now you can test the model in the synthesis notebook

can i get a link to the one im supposed to use

since the last training one was outdated, this synthesis one probably is too

Its in #841437191073955920

thank you!

Make sure to make your model public in google drive

So you can use it in synthesis

i have!

IT WORKS

HAHAHAHA

Yup, it’s actually alright and fine, as long as you can see one visible line in each graph, no matter if it cuts short or long or diagonal… But, if the pixels are scattered all around the place and becomes more “static” like, then there is something wrong with the training process (aka dataset issues)

awesome :D

Cuz it's dumb I convert with audacity too, it's just something it likes to do

Hi there there a notebook where you can give your ai personality mine kinda sounds monotone

The pronounce dictionary always messes it up

Nope You could try training a talknet model, because talknet accepts reference audio for copying the pitch and timing Or, you could tune your output from your current model in external software such as melodyne, newtone, or vocalshifter

Ok thank you

Also is there a way to make the prounouce dictionary functional?

Sorry for bad English

You mean arpabet input when synthesizing? You'll need to train on a notebook that enables arpabet The one called arpabet enabler in #841437191073955920 does — so does the pipeline one

Ok thank you I’m new to this so, training a tacotron model I already have with that will enable it?

Simply put

It will retrain the model from scratch