# machine-learning
- p
PUMPKINEATER
02/19/2022, 9:13 PMFeel free to translate what Joe Biden said - a
Alphabet
02/20/2022, 10:05 PMI have 3 hours worth of audio data, should I use tacotron or talknet to train for better results? Or does it not matter? - u
(Dawn) Will Draw Fictional Women
02/20/2022, 10:07 PMi would tacotron personally - u
(Dawn) Will Draw Fictional Women
02/20/2022, 10:07 PMbut both is always good - u
(Dawn) Will Draw Fictional Women
02/20/2022, 10:07 PMhell, training talknet can improve your tacotron - u
(Dawn) Will Draw Fictional Women
02/20/2022, 10:08 PMif you have a diverse set of emotions in your dataset i would also recommend training tacotron on the new pipeline - u
(Dawn) Will Draw Fictional Women
02/20/2022, 10:08 PMitd be cool to see some results from torchmoji - a
Alphabet
02/20/2022, 10:15 PMthanks for the tips. Almost done with the dataset, probably a couple more hours before I finish transcribing it all - h
hecko
02/20/2022, 10:38 PMout of curiosity what voice are you doing? - a
Alphabet
02/20/2022, 10:41 PMKayleigh McEnany - u
user
02/21/2022, 12:32 AMhey so im new to this and the synthesis notebook isnt working, does anyone know how to fix it? - u
user
02/21/2022, 12:33 AMbasically its saying my model is private when I made it public - m
mega b
02/21/2022, 12:46 AMMake a new cell, run
then restart runtime!pip install --upgrade gdown - m
mega b
02/21/2022, 12:46 AMAnd try again - u
user
02/21/2022, 3:13 AMthank you - u
user
02/21/2022, 4:50 AMuhh how do I fix this? - m
mega b
02/21/2022, 4:55 AMIt means your wav files in your dataset are not 22khz, mono - u
user
02/21/2022, 4:55 AMhuh weird - u
user
02/21/2022, 6:31 AMHow do I fix this? - u
user
02/21/2022, 6:31 AMAm I missing a line - u
user
02/21/2022, 8:17 AMI looked through the same problem people were having as me and I dont have any blank transcripts - u
user
02/22/2022, 3:59 AMThis is how it looks at about 190 epochs. I know it's supposed to be a line, but the synthesized voice isn't actually that bad. - a
a cat came into my house
02/22/2022, 4:15 AMwhat the fuck - u
user
02/22/2022, 5:01 AMHad the same problem - u
user
02/22/2022, 5:01 AMI fixed it by deleting the first wav file and erasing it from the txt file - u
user
02/22/2022, 4:35 PMHow many voicelines did you have? - u
user
02/22/2022, 5:30 PMI think about 150 - m
myposynator
02/22/2022, 6:17 PMHello, could someone help me? I've been trying to train my talknet model for weeks now, but I get an error when installing NeMo. It is saying unable to install numpy and pre-trained model. Does anybody know how to solve this? - u
(Dawn) Will Draw Fictional Women
02/22/2022, 6:21 PMgdown brokey - u
(Dawn) Will Draw Fictional Women
02/22/2022, 6:21 PMrun a cell with
and restart the runtime!pip install —upgrade gdown