# machine-learning
- s
strange
04/21/2022, 3:03 AMCan you please give notebook link sir - p
PUMPKINEATER
04/21/2022, 3:03 AMYou need to train a model first - p
PUMPKINEATER
04/21/2022, 3:03 AMThe models aren't preloaded - s
strange
04/21/2022, 3:03 AMAnd how do I do that - s
strange
04/21/2022, 3:04 AMtacotron 2? - p
PUMPKINEATER
04/21/2022, 3:04 AMMake a dataset first - p
PUMPKINEATER
04/21/2022, 3:04 AMzip the wavs seperately - s
strange
04/21/2022, 3:04 AMI have dataset - p
PUMPKINEATER
04/21/2022, 3:04 AMand make a transcription - s
strange
04/21/2022, 3:04 AMOk - p
PUMPKINEATER
04/21/2022, 3:04 AMOk. - p
PUMPKINEATER
04/21/2022, 3:04 AMZip all the wavs - p
PUMPKINEATER
04/21/2022, 3:05 AMin one zipped folder - p
PUMPKINEATER
04/21/2022, 3:05 AMand leave the transcript raw - s
strange
04/21/2022, 3:05 AMThanks if I get error can you please help me - s
strange
04/21/2022, 3:05 AM? - j
Justin
04/21/2022, 3:05 AMJust go through this guide once: https://awesomeface349.gitbook.io/uberduck-written-tutorial/ - s
strange
04/21/2022, 3:06 AMOk - s
strange
04/21/2022, 3:12 AMThe duration of wav file should be 10 sec? - p
PUMPKINEATER
04/21/2022, 3:13 AM1-12 each - j
Justin
04/21/2022, 3:17 AMI usually keep it under 6-8 secs - j
Justin
04/21/2022, 3:18 AMor 2-11 - j
Justin
04/21/2022, 3:18 AMmaybe 6-10 - u
(".Lukus_bt.")
04/21/2022, 3:22 AMtest voice - u
user
04/21/2022, 4:10 AMfazbear - m
mildrefusal
04/21/2022, 10:40 AMDo you use TalkNet to synthesise Tacotron2 models? - h
haru0l
04/21/2022, 12:52 PMno thats impossible as they're different architectures - c
coooooper
04/21/2022, 1:42 PMso what exactly does hifi gan do? - c
coooooper
04/21/2022, 1:42 PMthe fine tuning