# machine-learning
- t
TwoTimmm
10/23/2021, 2:30 PMWelp, the training failed - t
TwoTimmm
10/23/2021, 2:30 PMHis dataset is bad I guess - u
(Dawn) Will Draw Fictional Women
10/23/2021, 2:46 PMi think messing with the batch size and learning rate might get em working - t
TwoTimmm
10/23/2021, 2:49 PMThis is the current result - t
TwoTimmm
10/23/2021, 2:49 PMHe was supposed to say "You lack imagination." - t
TwoTimmm
10/23/2021, 2:49 PMCamille was less of a pain, she literally was clean and great before her Validation loss even dropped to 0.09 - t
TwoTimmm
10/23/2021, 2:53 PMMaybe it could be because his dataset includes him saying stuff half-whispering and long like this: - t
TwoTimmm
10/23/2021, 2:53 PMOr this - t
TwoTimmm
10/24/2021, 9:44 AMnow that I thought about it - t
TwoTimmm
10/24/2021, 9:44 AMhow do you make a screaming or a laughing model? - u
(Edd) bruh moment
10/24/2021, 9:46 AMI do make screaming or laughing models of course - t
TwoTimmm
10/24/2021, 9:47 AMNo I meant - t
TwoTimmm
10/24/2021, 9:47 AMHow exactly do you do it - t
TwoTimmm
10/24/2021, 9:47 AMdo you just put bare laughing clips and transcribe them as "hahaha" etc. or do you need the line to include both laughing and normal talking so it doesn't break? - u
(Edd) bruh moment
10/24/2021, 9:48 AMYes - u
(Edd) bruh moment
10/24/2021, 9:48 AMI mean I put them as "Ha ha ha ha ha" - t
TwoTimmm
10/24/2021, 9:49 AMso it's just bare laughing clips? - u
(Edd) bruh moment
10/24/2021, 9:49 AMye - t
TwoTimmm
10/24/2021, 9:50 AMk - t
TwoTimmm
10/24/2021, 10:23 AMmaking a riven model that possibly can say ugh - t
TwoTimmm
10/24/2021, 10:43 AMwhat? - u
(Edd) bruh moment
10/24/2021, 10:50 AMuse this to help you out: https://awesomeface349.gitbook.io/uberduck-written-tutorial/ - h
hecko
10/24/2021, 10:50 AMyour audios are in stereo - t
TwoTimmm
10/24/2021, 10:50 AMI converted them all to mono - h
hecko
10/24/2021, 10:50 AMor at least one of the audios - h
hecko
10/24/2021, 10:51 AMcheck the 11th or 12th wav - t
TwoTimmm
10/24/2021, 10:52 AMsays mono - t
TwoTimmm
10/24/2021, 10:55 AMso I don't think the problem is in mono - h
hecko
10/24/2021, 10:58 AMwell the input is twice as large as it should be - h
hecko
10/24/2021, 10:58 AMso maybe it's 44khz