# machine-learning
- s
SuperLuig
06/02/2021, 10:49 PMnvm - i
IBob012
06/02/2021, 10:50 PMhelth - s
SuperLuig
06/02/2021, 10:51 PMok well he's speaking nonsense but its a start - s
SuperLuig
06/02/2021, 10:52 PMhmm - s
SuperLuig
06/02/2021, 10:52 PM200 epochs, next target? - s
SuperLuig
06/02/2021, 10:56 PM166, dunno what the hell is going on with it - s
SuperLuig
06/02/2021, 10:56 PMthis shit doesnt look too good bro - s
SuperLuig
06/02/2021, 10:57 PMbut yeah i'll test it at 200 - s
SuperLuig
06/02/2021, 11:03 PMjesus CHRIST this is bad - s
SuperLuig
06/02/2021, 11:04 PM200 - p
pi
06/02/2021, 11:05 PMbecause the dataset is tiny? - s
SuperLuig
06/02/2021, 11:06 PMthe resort owner literally had 1:20 of data and turned out great - s
SuperLuig
06/02/2021, 11:06 PMsnugglemagne here has 71 lines and 2:36 of data - p
pi
06/02/2021, 11:06 PMthat's still very small - p
pi
06/02/2021, 11:07 PMit's also important how similar the voice is you want to clone to the librispeech voice - s
SuperLuig
06/02/2021, 11:08 PMFINALLY, RESULTS - s
SuperLuig
06/02/2021, 11:09 PMsounds good but he's speaking REALLY fast for some reason - s
SuperLuig
06/02/2021, 11:14 PMgonna try it with uberduck juice real quick - s
SuperLuig
06/02/2021, 11:19 PM:/ - s
SuperLuig
06/02/2021, 11:19 PMthis is like, actually fucked - s
SuperLuig
06/02/2021, 11:21 PMwhat the hell am I doing wrong here 😭 @User - s
SuperLuig
06/02/2021, 11:22 PMi've never had this many issues with a single model - s
SuperLuig
06/02/2021, 11:23 PMnothing is wrong with my dataset - s
SuperLuig
06/02/2021, 11:25 PMwait, hold up - s
SuperLuig
06/02/2021, 11:25 PMthat seems to only affect the hifi notebook - i
IBob012
06/02/2021, 11:26 PMyou turned off the pronunciation dictionary option right - s
SuperLuig
06/02/2021, 11:26 PMOH