Stanley Parable Narrator | Kevan Brighting

Create

Stanley Parable Narrator | Kevan Brighting

EnglishRVC V2
lusbert_ user image
lusbert_
1 year ago
👀

1k

👍

21

🪄

129

Description

16 batch size RMVPE 12 minute dataset ripped from game 40k pretrain 17000 steps < the infers are vocals ripped from Half Life 2 the transpose used is in the name of the audio files

Comments

resizable user image
resizable
1 year ago

Sometimes I wonder how lusbert is able to craft these good models

lusbert_ user image
lusbert_
1 year ago

50ms truncate silence

lusbert_ user image
lusbert_
1 year ago

and just overtrain since overtrainning doesnt exist for good datasets

resizable user image
resizable
1 year ago

Really?

lusbert_ user image
lusbert_
1 year ago

tensorboard is broken so i cant send the graph but not like it matters

lusbert_ user image
lusbert_
1 year ago

yea

resizable user image
resizable
1 year ago

Interesting

lusbert_ user image
lusbert_
1 year ago

Raven tested and apparently you should just overtrain to 1000

lusbert_ user image
lusbert_
1 year ago

500 stil has noises

lusbert_ user image
lusbert_
1 year ago

and 750 has engine noises

lusbert_ user image
lusbert_
1 year ago

also noise gate is bad down ward expand is better

lusbert_ user image
lusbert_
1 year ago

instantly cutting off a sentence basically is bad you need to cut when there isnt any setences being said

lusbert_ user image
lusbert_
1 year ago

<@622856015444049937> proof that you didnt need a 4 hour dataset nor tensorboard..

resizable user image
resizable
1 year ago

So what about your leviathan model

lusbert_ user image
lusbert_
1 year ago

that one was old processing so not as good as it could've been

lusbert_ user image
lusbert_
1 year ago

and mangio crepe is basically bad by now

lusbert_ user image
lusbert_
1 year ago

RMVPE with RMVPE infer best

resizable user image
resizable
1 year ago

Oh

resizable user image
resizable
1 year ago

Anyways good model

lusbert_ user image
lusbert_
1 year ago

real

lusbert_ user image
lusbert_
1 year ago

even Mode Collapses dont exist

lusbert_ user image
lusbert_
1 year ago

Raven had like 50 mode collapses in one model and it still sounded amazing

lusbert_ user image
lusbert_
1 year ago

literally the 1000th epoch was a mode collapse

lusbert_ user image
lusbert_
1 year ago

and it sound better even then

lusbert_ user image
lusbert_
1 year ago

bruh that was months ago he prob didnt know better but for GAN the stuff is correct

x_14106 user image
x_14106
1 year ago

that's what I said and someone booed me for it

lusbert_ user image
lusbert_
1 year ago

bruh

lusbert_ user image
lusbert_
1 year ago

ig you need to show proof that they cant understand :0

lusbert_ user image
lusbert_
1 year ago

like this

x_14106 user image
x_14106
1 year ago

I'm harvesting cuz I'm ballin

lusbert_ user image
lusbert_
1 year ago

Welcome To the Internet cover ong sounds fire

x_14106 user image
x_14106
1 year ago

hmmmmmm rhymes with aven tho

x_14106 user image
x_14106
1 year ago

but that's ok

x_14106 user image
x_14106
1 year ago

it was ages ago

lusbert_ user image
lusbert_
1 year ago

oh rip right i cant send that

lusbert_ user image
lusbert_
1 year ago

copyright

lusbert_ user image
lusbert_
1 year ago

well such a loss

lusbert_ user image
lusbert_
1 year ago

well just know that it sounds fire :0

lusbert_ user image
lusbert_
1 year ago

also holds up pretty well at the laughing part

lusbert_ user image
lusbert_
1 year ago

sounds like he's mocking me

fazemasta (not MJ) user image
fazemasta (not MJ)
1 year ago

😭

Trudy Is Back user image
Trudy Is Back
1 year ago

What

Add a comment

Samples

New
Classic
1. Singing
Male
English
2. Singing
Female
English
3. Singing (Dry)
Female
English
4. Singing (High)
Female
English
5. Singing 2
Male
English
6. Singing (Dry)
Male
English
7. Singing (Dry, High)
Male
English

Pitch

Selected Audio
Selected Audio