how long was the dataset in total cause you got a case of "abnormal sounding human" shitpost

look at my nails model

that aint a voice but it's not speaking (kinda)

try these: - truncating if not already - looping the same sound up to around ~2 minutes of data

yeah,

in theory having more data would train it to match the original sound

so i say just yoink all these in audacity and loop it till around 2 minutes

then optionally truncate it so that that *tiny* pause at the end doesnt occur

so that the model doesn't train on silence

not truncated enough

dw though i got you

thats what it should sound more like

or just use that idm just credit me

the more you listen to it

skeleton beating trap beat prod. vinnystop

anyways im currently almost done cooking a v3 nails model

This is endermen