Oh yeah that's why but the model was approved but it's not showing up on uberduck

No, if it was approved, it would've been made primary. That's how approval works.

Oh yeah

ok

"I'll help anyone interested on how to earn 50k_$80k_100k in just 72 hours from the crypto market. message me now for more information how to get started📥 +18053722289

i cant log in

i got a question

is this the best training notebook for making voice models? https://colab.research.google.com/drive/1x-sB6K4BDQm9dzEqPND8fQzTcvX4Z3Hh

Yes

i got another question

what is the correct directory for finalizing the voice model by using this training notebook? https://colab.research.google.com/drive/1x-sB6K4BDQm9dzEqPND8fQzTcvX4Z3Hh

the same directory that your unfinalized model is saved in

is the the project folder?

No

It's /content/drive/MyDrive/Tacotron_2

it gives me an error

That's odd

i am trying to do a count dracula voice model

as my very first voice model for uberduck

So what went wrong?

here is the error that i am talking about

i had done everything correctly

i hope you can help me and could try to fix this error

checkpoints directory : /content/drive/MyDrive/hifigan/dsp/ Loading '/content/drive/MyDrive/hifigan/dsp/g_00000000' Complete. Loading '/content/drive/MyDrive/hifigan/dsp/do_00000000' Complete. Traceback (most recent call last): File "train.py", line 284, in main() File "train.py", line 281, in main train(0, a, h, a.warm_start) File "train.py", line 59, in train steps = state_dict_do['steps'] + 1 KeyError: 'steps'

Got this when training hifigan.

i have wav reference audios and the transcript but i have no idea how to do anything else cuz the tutorials outdated i think

try this https://docs.google.com/document/d/1Gt_TNLAWyNzr0bjtYOE9Qm8yEvIL5yDq-W4E0kH5XW0/edit

I think my name and icon are missing from the Postal Dude voice model that I collaborated with another user to make?

anyone know if it's normal for sample inference files to sound like the ai is speaking through a fan

im up to my 100th epoch on my second trial run. last time i guess the sample equalizer notebook didn't upsample all of the files because some were still 20khz