# machine-learning
- w
WTLNetwork (❌ Voice requests)
05/07/2021, 6:32 PMAh, I think I might have sorted it. I factory-reset the runtime and set the batch size to 6 and it's just started chugging along. I think I may have made a rookie error and not properly reset the runtime before changing the batch size. - w
WTLNetwork (❌ Voice requests)
05/07/2021, 6:33 PMThanks for your help. 😄 - c
chickennugget
05/07/2021, 10:40 PMwhat's the minimum amount of time to get an ai voice sounding decent - k
Kosmikist
05/07/2021, 10:40 PMAround 5-10 minutes give or take - c
chickennugget
05/07/2021, 10:45 PMalright i think i'll be able to break that minimum time for annoying orange - c
Caster (Spooky M. AI)
05/07/2021, 10:48 PMhey chic - c
Caster (Spooky M. AI)
05/07/2021, 10:48 PMkeN - p
pi
05/07/2021, 11:08 PMi'd say 10 minutes is the minimum - t
The Watts and the Waves
05/07/2021, 11:30 PMthough every voice is different. I've been training 2 1/2 hours with... questionable results - s
SidPlays_144p
05/07/2021, 11:30 PMwho the - t
The Watts and the Waves
05/07/2021, 11:31 PMthis character was supposed to say "hello haha!" but uhhhhh he's taking a page from Adam's Freddie model at the moment 😂 - p
pi
05/08/2021, 12:04 AMstop training now - p
pi
05/08/2021, 12:04 AMit won't get better anymore - p
pi
05/08/2021, 12:04 AMit's highly likely that your dataset is too bad - o
OctolingVladTTS
05/08/2021, 12:05 AMcan I already use tacotron? - p
pi
05/08/2021, 12:06 AM? - o
OctolingVladTTS
05/08/2021, 12:06 AMI want to ask if I can use tacotron now - i
Isaac Rodriguez (Odcub)
05/08/2021, 12:08 AMwoah @User i didnt know your here too 😄 - o
OctolingVladTTS
05/08/2021, 12:09 AMYep - t
The Watts and the Waves
05/08/2021, 12:09 AMI appreciate the advice but I'm gonna continue anyway <3 I have samples I'm adding in anyway~ - p
pi
05/08/2021, 12:12 AMplease don't waste resources - p
pi
05/08/2021, 12:12 AMif it hasn't converged after 2.5h it won't converge at all - k
Kosmikist
05/08/2021, 12:14 AMYou should definitely train way less, it saves time and resources. Once the validation loss is at 0.1 thats when you know you should stop training - k
Kosmikist
05/08/2021, 12:15 AMI understand, do as you wish. I’m just giving advice from my experience is all - t
The Watts and the Waves
05/08/2021, 12:16 AMI super appreciate that and I'll definitely keep that in mind! - p
pi
05/08/2021, 12:16 AMyou're wasting resources - p
pi
05/08/2021, 12:16 AMit's certain that it wont converge anymore - p
pi
05/08/2021, 12:17 AMplease stop this asocial behavior - t
The Watts and the Waves
05/08/2021, 12:18 AMI'm simply asking for the unsolicited advice to stop, nothing asocial about maintaining boundaries :) - f
F.B
05/08/2021, 12:18 AMDon't say we warn you