# machine-learning
- a
Aniball
06/30/2021, 5:30 AMhere's what he sounds like with hifigan - a
Aniball
06/30/2021, 5:39 AMthis is what happens when you break him - a
Alexius08
06/30/2021, 7:03 AMUsing only over one and a half minute of source audio, I managed to make a model for George Patton. If anyone else is interested on this one, I'd submit it. - a
Alexius08
06/30/2021, 7:06 AM"I will wipe you the fuck out with precision the likes of which has never been seen before on this Earth, mark my fucking words." - m
M!nerman
06/30/2021, 4:05 PMInteresting stuff, the pitch rhythm matching. How does it work? - m
mega b
06/30/2021, 4:06 PMTalknet - c
Couch
06/30/2021, 4:07 PMwill there be a talknet training/synthesis notebook anytime soon - m
mega b
06/30/2021, 4:07 PMthere is one already - m
mega b
06/30/2021, 4:07 PMone sec - m
mega b
06/30/2021, 4:07 PM - c
Couch
06/30/2021, 4:07 PMnow that is what i call. an awesome - c
Couch
06/30/2021, 4:08 PMoh nice i dont need to change the dataset structure - j
Jara
06/30/2021, 5:00 PMupload this into the music part copy paste this: " I wanna be gay 'cause you're fucked and I wanna savior it But it's cool, after smashing you I'll Pay Pal you two kay " - a
Awesome Face (GTX 1660 Ti)
06/30/2021, 5:00 PMuh... - a
Awesome Face (GTX 1660 Ti)
06/30/2021, 5:00 PM#859823464847638538? - j
Jara
06/30/2021, 5:02 PM - j
Jetamex
06/30/2021, 5:08 PMso is the data set simular to tacotron? I have 100 audio sentences of my voice , do i just upload those? - m
MoarMosspato
06/30/2021, 5:08 PMtraining mephone4 (which was the first model I tried to make in the non-simpified training when uberduck existed) from inanimate insanity. The training looks surprisingly good with only 20 lines from him. - c
Couch
06/30/2021, 5:08 PMyes - j
Jetamex
06/30/2021, 5:08 PMok, thanks - c
Couch
06/30/2021, 5:09 PMpaths must be relative to the dataset archive - j
Jetamex
06/30/2021, 5:13 PMdoes this format work for my list? - c
Couch
06/30/2021, 5:17 PMthat should be good - j
Jetamex
06/30/2021, 5:19 PMok, sweet - c
Couch
06/30/2021, 5:21 PMtraining talknet and the (gen) loss is not changing at all it is stuck at 30 - c
Couch
06/30/2021, 5:21 PMnow that is what i call. a sus moment - j
Jetamex
06/30/2021, 5:22 PMvery sus - j
Jetamex
06/30/2021, 5:23 PMjust a few more questions, what do i do for these, i changed the list and dataset but no idea for this - c
Couch
06/30/2021, 5:25 PMi made val the same as training filelist output dir i just changed name_of_character to. the name of the character - c
Couch
06/30/2021, 5:25 PMi believe output dir is where checkpoints are saved