# machine-learning
- i
iGerman
08/14/2021, 7:41 AMi dont think it's right to distribute the model to anyone before asking for permission from the person i trained it on - m
mega b
08/14/2021, 7:42 AMoh i was just going to run it under hifi gan - m
mega b
08/14/2021, 7:43 AMup to you, anyway - i
iGerman
08/14/2021, 7:43 AMdoing that myself right now haha - m
mega b
08/14/2021, 7:43 AMooh okay lol - m
mega b
08/14/2021, 7:43 AMyou should also fine tune hifi gan - m
mega b
08/14/2021, 7:43 AMmake it sound pitch perfect - i
iGerman
08/14/2021, 7:43 AMhavent done that yet, should definitely try - m
mega b
08/14/2021, 7:44 AMsweet, lemme know how it goes. notebook -> #841437191073955920 - i
iGerman
08/14/2021, 8:08 AM250 epochs on papa's model, for now, not finetuned yet - m
matty1406
08/14/2021, 2:36 PMthis is my mrs pomp model, it only has 18 audio files but it sounds pretty surprisingly ok minus the little seizure at the end... - j
JacobLenstar
08/14/2021, 4:06 PMSaddam just does not want to train - j
JacobLenstar
08/14/2021, 4:06 PM82 full sentences with little to no background noise - f
FrankZ
08/14/2021, 5:52 PMjust making a dataset of Maurice from Madagascar - f
FrankZ
08/14/2021, 5:53 PMM O R I S - f
FrankZ
08/14/2021, 5:54 PMPUT THE FUNNY MOMENTS ON THE SCREEN - m
matty1406
08/14/2021, 5:56 PMi have fine tuned the mrs pomp model (dont know if I did it properly because I cant see much of a difference... - u
user
08/14/2021, 6:05 PM@User i have 0.wav but for some reason it thinks i dont - s
Scramble
08/14/2021, 6:09 PMOk so I'm having problems, I am training the furby dataset and it keeps going good only to go to shit - s
Scramble
08/14/2021, 6:09 PMand this as well - u
user
08/14/2021, 6:18 PMfixed - u
user
08/14/2021, 6:53 PMi got this, but then my spectro grams changed after, can i still stop the model? - u
user
08/14/2021, 9:34 PMincredible - h
Holos
08/15/2021, 1:09 AMDoes anyone normalize their audio here - u
user
08/15/2021, 1:09 AMuh - u
user
08/15/2021, 1:09 AMwhat - h
Holos
08/15/2021, 1:14 AMNormalizing their volume or amplitude in any way - s
Scramble
08/15/2021, 1:33 AMwhat - c
Couch
08/15/2021, 1:33 AMi use ffmpeg's loudnorm filter then normalize it on audacity - h
Holos
08/15/2021, 1:47 AMAh ok thanks I’m planning to use ffmpeg as well