# machine-learning
- k
Kosmikist
04/28/2021, 1:43 AMSounds like a good idea I'll keep it on the back burner for now - k
Kosmikist
04/28/2021, 1:44 AMIn the mean time I'll train Dio some more and maybe do a couple other characters. I have the audio files I need and its much clearer than monster so hopefully it'll go well - e
Echoes
04/28/2021, 1:45 AMI'm starting a dataset for Byakuya Togami from Danganronpa: Trigger Happy Havoc - p
PUMPKINEATER
04/28/2021, 1:46 AMlong legged dude - e
Echoes
04/28/2021, 1:47 AMpretty much lmao - t
Toasty
04/28/2021, 2:02 AMWorking on a new model - s
SidPlays_144p
04/28/2021, 2:03 AM - t
Toasty
04/28/2021, 2:03 AMTonight, on Bottom Gear I become a speech synthesis model - t
Toasty
04/28/2021, 2:10 AMHere's how it sounds so far - s
SidPlays_144p
04/28/2021, 2:26 AMThey're just giving advice - z
Zachary(retired f/ voice clonin)
04/28/2021, 12:06 PMJust got a trained Rocko Tacotron2 model fixed by Walt that got disconnected and it's mute - u
user
04/28/2021, 2:12 PMYou're being rude again... - o
OctolingVladTTS
04/28/2021, 4:03 PMforgiveness - o
OctolingVladTTS
04/28/2021, 4:03 PMis that sometimes I behave like this when I lose my patience when I get an error in the tts tacotron creation - w
Waltman13 (semi-active)
04/28/2021, 5:01 PMdid you add more wav files? like I said, you need about 40 audio clips in order for the voice to work. - o
OctolingVladTTS
04/28/2021, 5:02 PMthat's what i'm doing with luigi - w
Waltman13 (semi-active)
04/28/2021, 5:03 PMI was asking zachary. - o
OctolingVladTTS
04/28/2021, 5:03 PMOk Ok - f
F.B
04/28/2021, 5:36 PMSo, the Incredible polo model, I got a out of memory error - f
F.B
04/28/2021, 5:37 PMI thought it was a batch size problem, so I changed the batch size from 30 to 20 - f
F.B
04/28/2021, 5:37 PMThat still didn’t work - f
F.B
04/28/2021, 5:38 PMI checked the audio files and the Transcript to see if there is anything wrong with it - f
F.B
04/28/2021, 5:38 PMAnd it was fine - f
F.B
04/28/2021, 5:38 PMI used at least 54 audio files to make it possible - f
F.B
04/28/2021, 7:03 PMIs there an issue? - p
pi
04/28/2021, 7:08 PMuse fp16 - p
pi
04/28/2021, 7:09 PM - t
Toasty
04/28/2021, 8:16 PMHow long should I train my model if I have 2 hours of audio? - t
Toasty
04/28/2021, 8:16 PMThe dataset in total is 2 hours long, with 1506 files. - t
Toasty
04/28/2021, 8:16 PMI was thinking an hour?