#Cannot find the model in “Inferencing voice” after training
30 messages · Page 1 of 1 (latest)
Check the disk remain space
try using the minimun settings in step 3 while training
it fixed my issue
most of the time its too high of a batch size so lower it
run mine at 25 and it seems good. i have a 3080
rn i run 25, 200, 25
Okay. does to many voice files play a part in it. i have about 20 min of 10 sec cuts of my voice
I also have a RTX 2070 super
Same I have this as well
not too on that. I used 1 wav file that was 40 min. it seems to cut it up automatically. so i think youd be fine
Does Epoch amount determine The models speech pronunciation level My setting are Saving frequency (10) Total training epochs (20)
OK it just worked but my model sounds like it has an extreme lisp can’t pronounce certain words or do harmonize riffs, if you ever heard that song
Called hrs & hrs my model completely slaughters the first part thats a riff it’s terrible
Does Epoch amount determine The models speech pronunciation level My setting are Saving frequency (10) Total training epochs (20)
saw that lol
lol
im not 100% so dont take my word but from my inference its just more training
you could do like 50, 500, then ur batch
thatd take forever
Well if your right that's probably why my model sounds retarted "not enough time in the oven" lol
I might try this if i leave the house
lol fs
just make sure to set a good batch that ur gpu can handle or its all for nothing
Yeah, I'm still messing around with the value, evaluating much it can take