# machine-learning
- p
pi
05/31/2021, 8:48 PMlink to the dataset? - p
pi
05/31/2021, 8:48 PMinference with waveglow I assume? - s
Sonikkukokoro
05/31/2021, 8:50 PMEverything seemed fine according to the notebooks. No errors or anything. But yeah, here's the dataset. https://drive.google.com/file/d/1-3vEkXfoQe07momg0nUq_oaDUJ9KbU-n/view?usp=sharing - p
pi
05/31/2021, 8:50 PMthat's the model - s
Sonikkukokoro
05/31/2021, 8:50 PMOh. Do you mean all the voice clips and transcription? - p
pi
05/31/2021, 8:51 PMyes - s
Sonikkukokoro
05/31/2021, 8:51 PM - p
pi
05/31/2021, 9:00 PM@User The problem exists probably because almost all files are too short - p
pi
05/31/2021, 9:00 PMsee the faq for more info - p
pi
05/31/2021, 9:01 PMalso the game has much more audio for this character afaik - s
Sonikkukokoro
05/31/2021, 9:03 PMWell, it doesn't help that the Turret doesn't say a lot. - p
pi
05/31/2021, 9:06 PMyou're wrong I think - p
pi
05/31/2021, 9:06 PM - s
Sonikkukokoro
05/31/2021, 9:07 PMThat's GLaDOS. Not the Turret. You guys already have GLaDOS on the site. - s
Sonikkukokoro
05/31/2021, 9:08 PMThis is how many lines the Turret has. https://theportalwiki.com/wiki/Turret_voice_lines - p
pi
05/31/2021, 9:11 PMoh ok - s
SuperLuig
05/31/2021, 9:14 PM@pi not comparing, but didn't 15.ai have a turret voice that worked really well? - p
pi
05/31/2021, 9:14 PMI don't know - p
pi
05/31/2021, 9:15 PMmaybe? - s
SuperLuig
05/31/2021, 9:15 PMYeah there was - s
SuperLuig
05/31/2021, 9:15 PMI remember it - p
pi
05/31/2021, 9:15 PMok - p
pi
05/31/2021, 9:16 PMI assume the frequency range of the output is rather unfavorable - i
IBob012
05/31/2021, 9:16 PMit had and it was spot on - p
pi
05/31/2021, 9:16 PMand the length of the clips too - p
pi
05/31/2021, 9:16 PMso I'm really not surprised that it doesn't work well on vanilla tacotron - s
Sonikkukokoro
05/31/2021, 9:17 PMI put those lines in my dataset too.