download: https://huggingface.co/lyery/models/resolve/main/jeff_64e_1408s.zip?download=true
the other jeff model is overtrained asf so i made one that is not overtrained, i did this bc i like jeff, he's such a silly goober 
set index to 1 to make him sound exactly like he sounds in the game
index 0 allows him to speak/sing normally but imo he loses his charm
play with different index values, but i recommend just setting it to 1
model info 🤓 ☝️ :
dataset: 3:03 minutes
pretrain: klm 4 hifigan final version
batch size: 3
epochs: 64
f0: rmvpe
sample rate: 32k
steps: 1408
vocoder: hifigan
This model is compatible with mainline, older forks and w-okada voice changer
