#Michael Jackson (Dangerous Era Voice) (New) [RVCV2] [OG Pretrein] [150 Epochs]

26 messages · Page 1 of 1 (latest)

elder saddle
#

here is the voice of michael jackson About dangerous I hope you like it compared to the old one (or used original pretrein even if I noticed some defects and better than the old one in my opinion)

Link music voice Dangerous Era - Rock With You: http://sndup.net/8jrqv

Index raccometed 4.0 or 3,5 - 3,0

Picht raccometed +2 or 0 -1 for male fammale 0 or -1 or -2

train

Dataset 24:31
Rate 32k
Bucht size 7
Pretrein Original
Epochs 150

PS: credits for used voice 🙂 @elder saddle

Link voice: https://huggingface.co/Loren85/Michael-Jackson-dangerous-Era-Voice-v2/resolve/main/Michael Jackson dangerous Era V2.zip?download=true

coral juncoBOT
#
Rate the model

To rate this model, either use /rate or click the buttons below.

coral juncoBOT
dire token
#

nice but i think 32k is a super low rate, I'm pretty sure all the stems were at least 44k if not mostly 48k

#

idk how big the dataset is but I'd reckon the batch size is too low for it

#

epoch is too low and sounds undertrained

elder saddle
#

really?

dire token
#

not a great model but its cool

elder saddle
#

no offense

dire token
#

just need right settings and trained fully

elder saddle
#

type for example (because I use RVCV2 disco)

#

so 40k with 200 epochs for example?

dire token
#

thing is there is no one specific epoch all models can go under

#

u have to watch tensorboard graphs

#

and i bet 48k the best one you can do

elder saddle
#

ok

#

maybe I'll try to do it with 48k and 200 epochs to try

dire token
#

how big is dataset

#

or length

elder saddle
#

and written in the model description

#

@dire token

dire token
#

ah sorry

elder saddle
#

however, it is 24 minutes and 31 seconds

dire token
#

try batch size 8

elder saddle
#

ok

faint wharfBOT
#
Model Ready ✅

This model has been synced with Weights and is ready to use for free!