✨│ai-help

[prisma fern]

ight ill try this

alright i loaded the model. i got the acapella for the song i want to use

it needs a youtube link right

[frigid mural]

or google drive path

[rare goblet]

Ayo? @frigid mural level 1 !!!

[frigid mural]

idk

@half cove

I need helpp

[half cove]

with?

[frigid mural]

how do I like

Cover generated at /content/AICoverGen-NO-UI-en/song_output/PMV0pwRG8eE/sukuna's domain expansion be like🤣🤣 #jujutsukaisen (Gojo Satoru Ver).mp3
0

how do I see that

or open

or download

[half cove]

on the left

go on the colab file manager

and go to the corresponding path

[frigid mural]

ok

i'll try

YO IT WORKED

THANK YOU

[prisma fern]

HOW

my brain is actually fried after all this

im so close

[frigid mural]

why

what happened

[prisma fern]

python3: can't open file '/content/src/main.py': [Errno 2] No such file or directory
2

@title # Generate Cover

@markdown ## Output generated inside "FIX\song_output\random_number"

[frigid mural]

did it generate it

[prisma fern]

i have no idea man. im trynna make some MJ vocals to a Grover Washington song. i had the model of Mj and the acapellas of the song

[rare goblet]

Ayo? @prisma fern level 2 !!!

[prisma fern]

my brain just hurts seeing all this coding shit not work

[half cove]

did you install requirements before?

im guessing your using no-ui aicovergen

this is the guide https://aihubdocs.github.io/en/rvc/cloud/inference/aicovergen-no-ui/

[turbid pagoda]

Could anyone possibly help me. I am trying to import a hugging face voice model and whenever I do it comes up with a
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
and its every single model I try. Im not sure what else i can do did google ban more stuff?

im using google collab

[royal pond]

what is a nvidia driver??

[supple ibex]

what is your graphics card?

[royal pond]

what do you mean

[supple ibex]

what graphics card do you have in your computer

[prisma fern]

Cover generated at /content/AICoverGen-NO-UI-en/song_output/u1nsl6pVP9k/結城アイラ ⧸ Blessing（TVアニメ『聖女の魔力は万能です』OP主題歌）[Official Video] (Yuuri Ver).wav

i think mine worked

what now

[royal pond]

im on mobile phone

[supple ibex]

are you using colab?

[royal pond]

yeah

[supple ibex]

I think its broken right now

[royal pond]

yeahh i heard about that

[half cove]

try long install

[royal pond]

but i didnt know if it was true or not

[royal pond]

alr

[brittle wing]

can you guys send the best link to make ai covers pls

[royal pond]

im using one from a pinned message

should i send it?

[brittle wing]

sure

[royal pond]

alr!!

[half cove]

eh

[royal pond]

https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb#scrollTo=NEglTq6Ya9d0

[half cove]

thats not on pinned message but ok

[supple ibex]

can someone help me with this error?

[royal pond]

i saw it on a pinned message

.

[left oasis]

send me the current version of colab please

[half cove]

ohh

[half cove]

i mean it should work

[half cove]

-colab

[azure marsh]

Suggestions for @left oasis

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[half cove]

for ila rvc use the one on pins

[supple ibex]

can you help me with this error? I dont know why its happening

[left oasis]

I just have an error

[royal pond]

wellll, @supple ibex told me that colab was broken, but im gonna try to use the long install way to see if that works

[half cove]

show the error

[half cove]

i havent seen this before

[left oasis]

min

[half cove]

ask in #1192011222023950368 so more people see

[supple ibex]

ok thanks

[lean agate]

how can I use the thensorboard?
rn I'm training a model and I would like to see how the epoch are going
but there is no update

[left oasis]

i hate this

It started for me about 2 days ago

[royal pond]

colab its failing a damn lot

[lean agate]

rasp, sabes de crear modelos?

[half cove]

read dis or use the ila rvc on pinned #✨│ai-help message

@left oasis

also we have guides https://aihubdocs.github.io/en/rvc/cloud/inference/ilaria-rvc/

[royal pond]

nooo, me encantaría aprender vaya
sí es verdad que llegué a intentarlo, pero nada
todos mis modelos son descargados

[left oasis]

ok i tested

[royal pond]

bueno pero por lo que me han dicho ahora necesitas un archivo pth y index, meter las dos cosas en una carpeta con el nombre del personaje al que vayas a crear modelo de voz y convertirlo en zip (comprimirlo)
pero ni ideaaaa, no sé qué cosas tiene que contener cada archivo, así que no podría ayudarte😿😿

[left oasis]

And that's my second problem.

ok i tested new version

[half cove]

use new its better

[left oasis]

ok i testing

[half cove]

or just ila rvc

if your infering

[left oasis]

why are there so many buttons

[half cove]

the guide should explain it all

[left oasis]

and where is the guide I lost it

[half cove]

https://aihubdocs.github.io/en/rvc/cloud/inference/ilaria-rvc/

[left oasis]

stop training? I don't need to train my voices.

[prisma fern]

ive been tryna make a simple track for an hour

[polar plaza]

Bro how do you use the new one

[prisma fern]

nothing seems to work

Exception: The folder /content/AICoverGen-NO-UI-en/rvc_models/JustTwoOfUs does not exist.

i did all the steps

righ

right

[left oasis]

i found

[half cove]

nice

[left oasis]

poop master?

stop what AHJAHAHHAHAHA]

[left oasis]

you're not poop

[half cove]

bro

what

[left oasis]

your nickname

Why is he like this?

[half cove]

?

[left oasis]

?

[left oasis]

Are you real?

[spare shale]

@left oasis Что ты Хочешь он него?!

[left oasis]

это саппорт?

я просто интересовался почему он так себя назвал

[spare shale]

Это Его ник

[left oasis]

а не агрессией

[spare shale]

Я тоже

sweat

Он просто не понимает тебя Что ты Хочешь

[left oasis]

аа блин

хах

[left oasis]

о можно тебе тогда буду скидывать если что-то не получиться

[polar plaza]

How do you import a model on this

[spare shale]

По этому Пожалуйста не Спрашивай это

[left oasis]

да я не буду нормально все

[spare shale]

?

[left oasis]

ну например вдруг еррор у меня будет

[spare shale]

Если это colab Нет проблем

Но Локально я не Работаю

[left oasis]

хорошо

[half cove]

the the one on pins btw

[spare shale]

@half cove I think he won't ask Strange Questions anymore 😅

[left oasis]

Just forget it.

What did I write to you about

[spare shale]

Так же @left oasis Если тебе нужна русская Помощь ты Можешь написать #🌏│русский

[left oasis]

оо отлично спасибо

вы так всем помогаете?

[spare shale]

Да Я Общаюсь на всех Языках (Правда Translate Помогает 😅 )

[left oasis]

я немного не понимаю почему именно это еррор

[spare shale]

Ссылка не верна

[left oasis]

как так вроде все праавильно

[spare shale]

Но ошибка Пишет Другое

[left oasis]

я проверил даже закачка началась

[spare shale]

Можешь отправить ссылку сюда*

[left oasis]

хм странно все отлично хотя у меня её не было

извиняюсь

[spare shale]

??

[left oasis]

все хорошо она появилась

[spare shale]

😅

[left oasis]

модель

[spare shale]

Ну вот и Хорошо

[left oasis]

а здесь по моему надо дольше ждать

[polar plaza]

How do you import model on Ilaria

[spare shale]

You need a link to the Model, then click Download_model and paste the link there

[left oasis]

yes

[spare shale]

Наверное да

[polar plaza]

Yeah I tried it just told me that there was an error even tho I know the link still works 😭

[spare shale]

Can you send a link? so I can check

[half cove]

hes not downloading it right

it needs to be the index and pth in a zip

[spare shale]

He said the link is already working

[half cove]

not by itself

[left oasis]

omg thank you for helping

[rare goblet]

Ayo? @left oasis level 6 !!!

[half cove]

you need both to be in a zip file

[spare shale]

😮

Oh really, sorry, I didn’t see it (I asked for a link to check)

@left oasis в Zip 2 Файла?

pth и index?

[polar plaza]

I was using this https://huggingface.co/sail-rvc/alastor/blob/main/model.pth

[spare shale]

You must Create a ZIP inside the ZIP There must be model.pth and index

[weak hound]

Anyone know how to fix gaps in vocals? It seems like it's parts where the backing vocals overtake the main vocals.

[polar plaza]

ThT don’t make no sense WHAT DO YOU MEAN😭

[left oasis]

I had the same thing, but then she showed up there on her own

[polar plaza]

I’m confused

[left oasis]

да а что?

[tepid cave]

give shitmaster a break pleas

[left oasis]

ну pth и index по стандарту

[spare shale]

I mean that without index and zip it will not download

[polar plaza]

Will the easy GUI be fixed soon please tell me 👹

[half cove]

this is not a download link and also isnt correct

[left oasis]

саппорты вы лучшие если бы не вы то я не нашел

[half cove]

please read this guide https://aihubdocs.github.io/en/essentials/uploading-models-to-hugging-face/

[tepid cave]

ok ur banned

[polar plaza]

WHAT

[tepid cave]

!ban

/j

[spare shale]

https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw

[polar plaza]

This looks like it’s only to train??

[left oasis]

можно кое-что спросить а как ты сделал так что у тебя написанно когда день рождение?

[spare shale]

There you can create a cover

[polar plaza]

I’m using a model that was already made wtf

[polar plaza]

Huh-

[spare shale]

It's EasyGUI but without Interface

oh wait

But it only works if you have Colab pro 😅

[polar plaza]

I’m gonna cry

[left oasis]

аххахаха ладно 😁

[half cove]

it should work for free users btw

[left oasis]

don't cry

[spare shale]

He doesn't mean Train the Model, but create Cover
This Colab does not have an Interface for Creating cover only Model Checking

[polar plaza]

😭

[rare goblet]

Ayo? @polar plaza level 2 !!!

[half cove]

1. it does
2. the ui should work for free users as he changed the name of app.py to demo.py making it not be blocked by colab

[spare shale]

But there is also an Interface for creating Cover, but you need Colab Pro

[half cove]

yes and i talked with the person who made this

it might disconnect you but it should still work for most people

in any case for inference just use ila rvc

[polar plaza]

What do I do for literally everything else, again I’m using a model I got from huggingface, I did not make it. But everything else is just for training a model and I don’t want to do that
I’m so lost

[spare shale]

EEEEEH

[polar plaza]

?

[spare shale]

Can you DM me the pth and index?

[half cove]

ok so basically the model you are using the user did not set up the files correctly and did it separately instead of all in one zip file

download both

put it in a zip file

reupload it

boom

[spare shale]

Yes I will do it now

If he sends me files

[polar plaza]

Well thing is um I’m using multiple models

Do you want me to send the files for all?
I can separate them

[half cove]

depends if they are setup correct or not

[polar plaza]

Bro these don’t have index wth 💀

[spare shale]

The model will not work 😅

yea @half cove?

[half cove]

model will work without index

[spare shale]

Or am I misunderstanding again?

[half cove]

index is for accents but you can infer without it

[spare shale]

Cry

[polar plaza]

can someone plz fix the easy gui then because that’s literally the only one I’ve been able to use

And it’s BROKEN

[spare shale]

Which one did you use?

[polar plaza]

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing

This one

[spare shale]

Oh

[polar plaza]

Everytime I import a model it gives an error. It was working just fine a few days ago

[half cove]

ahh fuck it ill just fix the fucking downloader

[rare goblet]

Ayo? @royal pond level 3 !!!

[polar plaza]

BRUHHH LMAO

[royal pond]

i waited 18 fucking mins for this

[spare shale]

It's not that much

[half cove]

are you using t4 runtime?

[half cove]

for the code for the downloader

use thise

[royal pond]

i have been trying since 2 hours ago

[royal pond]

yesss

[spare shale]

@polar plaza Try using another colab -> https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing

[half cove]

then i dont know

[polar plaza]

This is the one I was using that didn’t let me download a model 💀

[royal pond]

damn😿

[polar plaza]

What is this?

[spare shale]

@half cove Is it possible to upload to https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing Without index?

[royal pond]

everythings right, idk what is happening to google colabs

[half cove]

the downloader but fixed

[half cove]

it should be

let me check

[spare shale]

@polar plaza Wait

[polar plaza]

?

[spare shale]

Wait @half cove check how to do it and reply in a few minutes

[warm snow]

hi, is there any way to run the w-okada rvc on google colab atm?

[spare shale]

Yes

https://colab.research.google.com/github/w-okada/voice-changer/blob/master/Realtime_Voice_Changer_on_Colab.ipynb#scrollTo=86wTFmqsNMnD

[warm snow]

I've tried that but I get disconnected bc of TOS all the time

is there a bypass or something for that?

[polar plaza]

I’m so confused

[spare shale]

Oh really?

[warm snow]

I'll try that thank you!

[spare shale]

I've been using it for more than 1 hour, everything works well

[half cove]

use the hina one

[royal pond]

i think its just broken like @supple ibex said

[half cove]

https://colab.research.google.com/github/hinabl/voice-changer-colab/blob/master/Hina_Modified_Realtime_Voice_Changer_on_Colab.ipynb

[polar plaza]

Is the code you just sent me on the actual website tho💀🙏🏻

[spare shale]

Okay np

Thx ❤️

[half cove]

it does work

[spare shale]

Can you tell how?

Need to use zip only with pth?

https://media.discordapp.net/attachments/1167948353833816154/1194044119706767500/gif_make_mikus_18.gif?ex=65aeeb3e&is=659c763e&hm=ab3f85fa1a1e7c023fb7d5569a94bbe1cdbabff67495f9f4aec517f0e1a236c2&

[polar plaza]

Bruh it’s still broken

[rare goblet]

Ayo? @polar plaza level 3 !!!

[polar plaza]

Cries and dies

[half cove]

ok

send me the link

let me check if its correct

[spare shale]

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[polar plaza]

☹️

Lmao

[brittle wing]

In this link, do you need to put the image link in the "icon_image" option?

[spare shale]

@polar plaza What is your Language?

[polar plaza]

English why

[half cove]

just use ila rvc

[spare shale]

Then why don't you understand? @~@

[half cove]

no

[brittle wing]

oh ok

[polar plaza]

I’m slow 😭🙏🏻

[crystal thunder]

Hello i have problem

[half cove]

if you leave it the same it should put one for you

[half cove]

read the ila rvc guide

[polar plaza]

IT DOESNT WORK FOR ME 😭

[half cove]

and use the link in pinned messages

[polar plaza]

bruh

[half cove]

what part

the download?

send the link

[polar plaza]

Yes

[half cove]

send the link your using

[polar plaza]

I decided to use this one instead lmao https://huggingface.co/Eylanchan/RottmntLeo/blob/main/ROTTMNT leo.zip

[half cove]

change blob in the link to resolve

right now thats not a download link

[polar plaza]

Will it still work even if it’s not my link??

[half cove]

wait

[polar plaza]

💀

[spare shale]

yea

@polar plaza try: https://huggingface.co/Eylanchan/RottmntLeo/resolve/main/ROTTMNT leo.zip

[half cove]

yea

[spare shale]

cry

If you don't understand this, I'm sorry, I won't be able to Help you.

[polar plaza]

I got it I think

[spare shale]

YAY!

Camera

[spare shale]

I remember your words

[polar plaza]

I will try to see if it works

[crystal thunder]

hello, it runs into an error when i try to download a model

[spare shale]

Send a screenshot

[crystal thunder]

doesnt let me

idk why tho

[spare shale]

If it's Local then I'm sorry I can't help

[half cove]

do what he said and screenshot

[crystal thunder]

no no

[spare shale]

Send to DM

[crystal thunder]

i have the screenshot

[spare shale]

Which colab are you using?

[half cove]

hes using

the

old

easygui

[crystal thunder]

damn

[rare goblet]

Ayo? @crystal thunder level 1 !!!

[half cove]

just use ila rvc on pinned messages

also now you should have perms

[crystal thunder]

o

k

[spare shale]

@crystal thunder Use this colab -> https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Ilaria_RVC_Tempfix.ipynb

[half cove]

now rn

@crystal thunder use the one on pinned

[half cove]

right now this is broken

we have to wait until hina wakes up

[spare shale]

Oh I'm sorry

[half cove]

should be fixed tomorrow

the one on pinned messages is the fix

[spare shale]

Okay
@crystal thunder Use this -> https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Ilaria_RVC_Tempfix.ipynb

[crystal thunder]

thanks

[polar plaza]

So It worked and said it downloaded

And then I went back to interference page and I put in the audio I was using and it gave me another error

What does that mean

[spare shale]

Screenshot

[polar plaza]

Oh nvm

Me thinks it worked

[spare shale]

Cat

[polar plaza]

It did work omfg

LOL

[glad zealot]

Ye I was thinking that could be the problem, will change in 12 hours

[polar plaza]

When you realize it was less complicated than you thought it would be

[spare shale]

Congratulations

[half cove]

[brittle wing]

How am I going to put the model in llaria RVC? There is no option to put the model link to do the AI

[spare shale]

You must Save the ZIP File your Voice (pth and index) Save and then use ilaria RVC

[half cove]

on the tabs it should have one for download model

at the top of gradio

[void salmon]

Hey ! Having the same issue here ^^ How do I input the model ? I download the .zip, alright, and then ? Sorry to ask I must be the milionth

[spare shale]

Are you using Local or Colab?

[half cove]

you need to upload it to huggingface and use the link

unless you are local

i can tell you how

[void salmon]

Colab. I have the huggingface link, but in Ilaria, I don't know where to put it 🙂

[rare goblet]

Ayo? @void salmon level 1 !!!

[onyx ferry]

[spare shale]

download_model

[half cove]

it should show here

[void salmon]

Thaaaaaanks you're the best !

Cannot use my eyes for the love of me

[spare shale]

sweat

okay.. I'll probably go do SD. Good luck to everyone!

[thin stump]

Should I remove this deep voice from the dataset? It's one of those imitations that you can do with your voice

[rare goblet]

Ayo? @thin stump level 13 !!!

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[long pike]

To anybody who can answer this, for weights.gg do you copy the model link and then paste it in the colab for rvc2 or do you download it then paste it in the colab?

[half cove]

for weights.gg ai voice models you need to reupload it

[thin stump]

Answering my own post #1201368092358492210 message

[dawn nebula]

keep that in ig

[prisma fern]

NVDIA problem

rn

[carmine charm]

hey guys

I'm having an issue with a Thanos Voice

it's coming out pretty robotic and i'm not sure if its the model or if im doing something wrong

could someone give me a handd

[spare shale]

Are you talking about Voice changer?

[carmine charm]

I'm using RVC

[spare shale]

colab or local?

[carmine charm]

not sure what that means, but local I imagine?

Harvest

[spare shale]

colab = Google / Browser
local = your PC / Laptop

[carmine charm]

Local

[half cove]

Don't use harvest

Use rmvpe

[carmine charm]

where's that?

[half cove]

Are you using a old fork like rvc gui?

[carmine charm]

Yes 😭

[half cove]

[spare shale]

https://tenor.com/view/kitty-eat-lava-minecraft-kitty-kitten-lava-gif-12699880907047808909

[half cove]

Try using mainline

#1159513888199540817 look at pinned then the website

It should have a guide on it

[carmine charm]

I can't find it

[half cove]

Too lazy to get the page for you cuz it's 1am and I'm on mobile but the website has a search bar and on the left it has dropdowns

[half cove]

Should be the top one

On guides

[carmine charm]

https://aihubdocs.github.io/en/

this?

[half cove]

Yes

[carmine charm]

So I'm not trying to make an AI cover, but just trying to do dialoge

[rare goblet]

Ayo? @carmine charm level 2 !!!

[carmine charm]

should i still click on the AI cover one

[half cove]

Like tts?

[carmine charm]

No

[half cove]

Or

[carmine charm]

like I speak

and convert it

i imagine that would sound less dead inside than tts

[half cove]

Real-time or are you just talking about converting normal speaking

[half cove]

Would sound monotone

[carmine charm]

not real time lmao

[half cove]

Then yea

Mainline

[carmine charm]

so i should just do the ai cover one

right?

[half cove]

It should have it all in the guide

Ask if you have any trouble

[carmine charm]

is it better to do local or cloud in your opinion

[half cove]

Local

[carmine charm]

makes sense

[half cove]

You don't have any restrictions

[carmine charm]

okay i'll try to figure this out on my own

thanks

[spare shale]

[carmine charm]

@half cove mainline is good for both coverting dialogue and covers, right?

[half cove]

Yes

[carmine charm]

[glad saddle]

Hi

I need help

[thin stump]

alright thanks

[spare shale]

Use this colab: https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Ilaria_RVC_Tempfix.ipynb

[loud talon]

Does anyone know why this error appears when I click generate cover in CoverGen_NO_UI_v2_en.ipynb?

[spare shale]

Hmm.. Can you show the full screen?

[gentle mural]

this is the only one that works for me none of the others do

[carmine charm]

@half cove

I downloaded it, and it said for me to open the zip

but it isn't a zip file

[half cove]

Where is you download it from

Did*

[carmine charm]

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006Nvidia.7z

[half cove]

Autocorrect shitting on me

[half cove]

7z file

[half cove]

Use 7zip

[carmine charm]

ohhh

💀

damn that was common sense

i tried using winrar

[half cove]

Other things will mess up when unzipping

Winrar and base

[glad saddle]

Got this problem on step 2

[dawn nebula]

which colab

easygui?

[glad saddle]

Yup

[half cove]

It's the only kalomaze "improved download one "

Fuck autocorrect actually shitting on me

[half cove]

Just use the ila rvc that's on pinned messages

[dawn nebula]

outdated

yeah use ila rvc

[half cove]

Btw the main ila rvc rmvpe is fucked so you gotta use the one in pinned messages untill tomorrow when hina changes it

[loud talon]

Like this?

[carmine charm]

7zip isnt working

😢

[half cove]

Huh

[glad saddle]

Shit, well

Thanks

[half cove]

[dawn nebula]

alr ty

[half cove]

What do you mean by not working

[carmine charm]

I installed it but it won't let me open the file with 7zip

now i have to reinstall the file and it'll take another 30 minutes

[half cove]

Use 7zip to unzip the file

[carmine charm]

I tried

it broke it

[half cove]

How du fuk

[carmine charm]

no idea

💀

maybe 7zip isnt installed correctly?

idk what im doing wrong

[loud talon]

Ok thanks, it was working fine for me a few minutes ago, I restarted the page, tried to use it again and out of nowhere that error occurred.

[rare goblet]

Ayo? @loud talon level 2 !!!

[glad saddle]

Another stupid question but... how do I add a voice model in Ilaria?

Sorry

[half cove]

At the top of gradio it was a tab called download models

[carmine charm]

im really confused why 7zip isnt working

maybe i didnt install correctly

[half cove]

I dont see how

[carmine charm]

but it sucks since it broke the file so i gotta wait another 30 minutes for it to download now

[glad saddle]

Oh, yeah, I'm an idiot

[rare goblet]

Ayo? @glad saddle level 4 !!!

[glad saddle]

Sorry

[carmine charm]

[glad saddle]

And thanks

[loud talon]

Ok, how do I do it?

I have already restarted the page and it still appears

Thanks, but I think the error is due to this, what does it mean?

Do I have to pay to continue using it?

[molten pecan]

Nah you can switch to another acc

Or wait 12 hours

[loud talon]

Oh I see, I'll try another account or wait those 12 or 24 hours, thanks.

Yeah it worked, thanks guys 👍 👍

[odd shale]

I really prefer using Ilaria RVC colab rather than AICoverGen and EasyGui

[loud talon]

I don't know, I tried about 5 and none of them worked for me, I use this one because it's the only one that has worked for me lol.

[odd shale]

5?

[loud talon]

Maybe I'll try to use that Ilaria RVC later, if you think it is better.

[glad zealot]

Agreed if only I'm not lazy to clean the audio

[odd shale]

All you need is an lead acapella and the instrumental.

[odd shale]

Installation only takes 2 mins.

[loud talon]

Well I don't remember well, I spent about 5 hours trying several and none of them worked lol.

[odd shale]

Maybe because you didn't know how to use them properly due to lack of knowledge

[glad zealot]

Me being lazy :D

[loud talon]

I followed the tutorials well, but I've seen that other people also get those same errors, so I assumed they didn't work.

[finite galleon]

Relatable

But at the end of the day inference is just inference

No quality differences between all versions

[thin stump]

Should I remove this voice crack from the dataset? I haven't heard a va scream like a banshee before

[carmine charm]

Can someone help me with 7zip

I feel like I'm doing something wrong

I can't get 7zip to open the RVC file

[rare goblet]

Ayo? @carmine charm level 3 !!!

[carmine charm]

Does anyone have any advice

i dont wanna ping anyone since i feel bad pinging people

but ive been trying to get this to work for hours

😢

[mild hornet]

hey, i am getting very cranky voice while using the hina modified w-okada voice changer via google colab. please help?

[desert niche]

-local

[azure marsh]

🖥️ Local stuff

• 🍏 Applio, by IA Hispano GitHub
• Mangio-RVC-Fork, by Mangio621 Huggingface
• RVC Studio, by SayanoAI Huggingface
• AICoverGen, by SociallyIneptWeeb GitHub
• Replay, by Replay Team Website
• Original RVC, by the RVC-Project team GitHub
• GPT-SoVITS, by RVC-Boss GitHub

Need some help? 🤔

You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.

More commands: -audio, -colabs, -overtrain, -realtime, -rvc, -help

[carmine charm]

Hey guys i'm getting an error code

"The system cannot find the path specified."

does anyone know why?

it happens whenever i do this

[young halo]

what can i do if i stop a training and then i want to resume it?

[jaunty flame]

help

@dawn nebula @tired kraken

[carmine charm]

bro they are streaming on twitch

[jaunty flame]

im using Applio

[carmine charm]

does anyone know why this is happening

ive been trying to figure it out for hours

[jaunty flame]

this?

[loud talon]

Hey guys do you have a tutorial to download and use Ilaria RVC?

[dawn nebula]

lemme read

[finite galleon]

Noisereduce??

When did applio add that

[dawn nebula]

did u extract the whole folder

[carmine charm]

does anyone maybe know why the go-web.bat for mainline RVC is not working

[carmine charm]

i thought so

how exactly do you do that

[dawn nebula]

@jaunty flame use mangio crepe

[carmine charm]

maybe im doing it wrong

[dawn nebula]

i think rmvpe is down or sth

[jaunty flame]

can i have the link please

[dawn nebula]

theres no mangio crepe here, try using crepe

[jaunty flame]

there is nocie reduce

[dawn nebula]

how did u extract

[carmine charm]

Let me send a screenshot because im very confused to be honest

[dawn nebula]

yeah sure

did you extract the whole folder, or just the file

[finite galleon]

Try making a new code cell paste this and run it

!pip install noisereduce

[jaunty flame]

i did but still error

[carmine charm]

@dawn nebula

[finite galleon]

Nah its on everything

[dawn nebula]

show me ur whole screen

[finite galleon]

Its just that noiseresuce inst installed

[carmine charm]

damn im gonna dox myself lmao

let me just blur out my name and stuff

[finite galleon]

You need to do !pip install noisereduce pretty sure

[dawn nebula]

yeah

[dawn nebula]

sure

[livid frigate]

What rvc is good for blue archive?

Or like

Just rvc v2?

[finite galleon]

Yeah applio did a commit that added noisereduce 3 hours ago

[dawn nebula]

click extract

[carmine charm]

I did

[dawn nebula]

yeah

[finite galleon]

You need to manually install noisereduce on the colab

[carmine charm]

it never does anything

[livid frigate]

Oh alright

[carmine charm]

i'll try again

[livid frigate]

Thanks tho

[jaunty flame]

just did but still

error

[finite galleon]

Is it a different error

[finite galleon]

Like any difference at the whole error?

[thin stump]

Is melband reformer a denoiser? The vocals on my dataset has no instrumentals

[carmine charm]

@dawn nebula

I clicked extract, what would the file be called after extracting?

[dawn nebula]

cleanup denoiser

[carmine charm]

im trying to find it

[dawn nebula]

most likely RVC1006Nvidia

[finite galleon]

If installing noisereduce didnt work i recommend using something else rather than applio rn

[jaunty flame]

any suggestions?

[carmine charm]

what type of file would it be?

[finite galleon]

Applio is currently being updated

[dawn nebula]

folder

[carmine charm]

it says its extracting to downloads but i never can find it

[finite galleon]

EasyGUI or IlariaRVC

[carmine charm]

yeah it does not exist

😭

[jaunty flame]

alright thanks

[carmine charm]

i dont understand what im doing wrong

ive extracted so many times

[thin stump]

ty, whoever owns this website really wants you to register, man

[dawn nebula]

if you log in you can basically having no wait times

[carmine charm]

Is there any reason why it wouldnt be working

ive been trying to figure this out all day

im really confused why this isnt working

[thin stump]

What's the duration limit for mvsep?

[young halo]

What if I input 2000 Epochs for my model and then I stop the training at 300 but I want to resume the training later?

[dawn nebula]

you retrain it using the same exp name and settings

it will continue

[young halo]

Will it continue from where I left it?

[dawn nebula]

yes

[dawn nebula]

i have no idea

[young halo]

Oh, do I need to save any of those checkpoint before?

On my Drive

[thin stump]

nevermind, i'll have to split my dataset into three parts then rejoin. Mr Modz said you can't do half an hour

[carmine charm]

wait there are two extract buttons

i think i might have been pressing the wrong one

Let me see if this solves it

@dawn nebula

it's a folder now

but i dont know what to do now

wait

[livid frigate]

How do we know how much pitch we gonna use to make the character sound like themselves?

[rare goblet]

Ayo? @livid frigate level 1 !!!

[neat totem]

hello i cant get any of my voices to work through discord, i was watching a video on youtube and i see that he has a output for the virtual audio cable that i dont have for some reason, you can see in the first image where his output is "Speakers (VB-Audio Cable)" but in my image you can see i dont have that option at all

[split latch]

what doe sthis mean

[naive relic]

ayo guys i want too use the voice model in this server in my python program. I got .pth file. How can i use this model in my python program?

[compact skiff]

So ive been testing the changer out in the settings on valorant and i feel like i can hear a cut once in a while when saying something. anyway to like stop that`?

[thin stump]

huge sigh. There's a lady talking in the background when the studio recording was going on

[tacit grail]

its still taking forever for stuff to render, why? i started a vocals only conversion on Ultimate Voice Remover and its only at 28% after an hour

[viscid moss]

I'm sure you're not using GPU

[tacit grail]

I actually am

[viscid moss]

hmmm

can u show me?

[viscid moss]

and... the main page 👀

[viscid moss]

How long is your audio? I know that the MDX23C needs a lot of resources but... 1 hour and it barely has 31% lol

[tacit grail]

2:58

[viscid moss]

[tacit grail]

2 minutes and 58 secxonds

[viscid moss]

I think something is not working correctly

[tacit grail]

thats what im thinking. like wtf

[viscid moss]

You can use it in google colab, with this colab I made:

[tacit grail]

[viscid moss]

https://colab.research.google.com/github/Eddycrack864/Music-Source-Separation-Universal-Colab/blob/main/Music_Source_Separation_Universal_Colab.ipynb#scrollTo=Mb7zrG7DSCle

[tacit grail]

i have no idea how to use that method

[oblique lava]

getting this error

JSONDecodeError Traceback (most recent call last)
<ipython-input-5-755770aa8734> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

this is what im using

https://colab.research.google.com/dri...

[austere ravine]

is there a tutorial for this

[viscid moss]

U need a google account, then in ur GDrive u need to create 2 folders. An input folder with the audios you want to separate and a folder where you want to save your separated audios. Then just run the first cell, and in the second cell replace "Separar" y "Vocales" for ur folder names. And that's it

[brittle wing]

the drive link is wrong

🐧

[viscid moss]

Outdated colab

[brittle wing]

el eddy se las sabia todas wtf

[viscid moss]

Es el EasyGUI de Kalo, eso es de Julio y ya se murió

[oblique lava]

where can i find the updated one

[rare goblet]

Ayo? @oblique lava level 1 !!!

[viscid moss]

-colab

[azure marsh]

Suggestions for @oblique lava

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[viscid moss]

guides on:
https://aihubdocs.github.io/en

[tacit grail]

what kind of colab doesnt allow a kits.ai link wtf

so i cant use a voice model from there?

[oblique lava]

yea guys im lost

[viscid moss]

Nope, Kits.ai models can only be used on kits.ai page

[tacit grail]

so this colab page is useless for me then

how do i make UVR render at a reasonable speed...

38% rn

[brittle wing]

with a good graphics card or a webpage

[tacit grail]

ok... let me correct myself

How do i make it where it renders it at the speed it is suppose to go?

it use to render pretty quickly but now it doesnt

[brittle wing]

depends on the audio

and model

[tacit grail]

all im doing is seperating the vocals into different vocal types

[viscid moss]

I mean, you can separate the vocals with colab and then put the separated audio in Kits.ai

Instead u can use MVSEP page

[tacit grail]

does it give me "lead_only_vocals"?

[viscid moss]

You can download the audio you want after separating, in this case: Vocals and Instrumental

[tacit grail]

omg... 50+ in queue

[tacit grail]

two and a half hours just splitting the vocals

thats just the vocals too. Now for another 2.5 hours for the instrumentals

[astral sparrow]

Anyone know why everytime I click train feature index it does not work AMD grahpic card

Does RVC 2 support AMD window 11 training voice

[naive relic]

ayo guys i want too use the voice model in this server in my python program. I got .pth file. How can i use this model in my python program?

[sand iris]

no

Sadge

[sand iris]

the existing code is kinda bad, your choices are unless to reimplement the thing (good luck lol), or use the command line interface from python

which won't really be all that good because you have to constantly load/unload models every time you generate audio

[astral sparrow]

So RVC and RVC 2 does not support AMD and window 11 for training right?

[thin stump]

soo melband reformer doesn't accept 9 minutes, great

[sand iris]

Yeah, it doesn't on windows 11 currently

[astral sparrow]

So there's no way for me to clone voice unless I pay in other app?

[rare goblet]

Ayo? @astral sparrow level 1 !!!

[sand iris]

There's always google colab

Which is free

[astral sparrow]

oh

I'll try

[sand iris]

I'm not really up to date on that front

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[sand iris]

RVC disconnected or EasyGUI should be the training codes

[livid frigate]

What is the pitch for blue archive character. Or i just have to guess it?

[sand iris]

Mostly guessing depending on your input voice lol

[livid frigate]

Oh...

Like ngl it's kinda hard to find it

[astral sparrow]

after cloning the voice that I want. Will it be a pth so I can put it on weights to use it

[livid frigate]

Eh pitch 15 is good

[sand iris]

yeah

[livid frigate]

I'll guess I'll just go with pitch 15

[feral marsh]

uhhh gpu isnt being detected.. any tips?

[forest sluice]

Colab or Local? If local, what's your GPU?

[feral marsh]

AMD Ryzen 9 3900X, not sure colab or local i just installed this today..

not really sure how it works yet

[copper nova]

that is a cpu sir

[gentle mural]

this is the only voice that works someone please help 😭 🙏

[feral marsh]

shit, AMD Radeon RX 6700

mb

[naive relic]

welp, you mean that i should develop my own inference in my program? damn

[sand iris]

You can take the existing code

...but yeah it's messy lol

[naive relic]

huh that's too bad

but i can try it. Right?

[sand iris]

Oh for sure

[naive relic]

ah ty

[sand iris]

Just gotta untangle the current code that's all over the place

If you want a rough idea on how things work :

they use a model called hubert from facebook's fairseq python library. This takes raw audio waveform resampled to 16KHz as input, and gives multiple arrays of length 786 as output

then there's also a pitch extraction model / algorithm that takes some kind of audio as input (depending on which), and outputs a series of numbers

those two data then get fed to the RVC ".pth" model

[naive relic]

uhhh wow

[sand iris]

yeah it ain't pretty

[naive relic]

looks very hard.. but i'll try..

ty bro

I guess I'll just have to develop a method to crawl the web UI and extract real-time audio output, lol.

[sand iris]

wait you need real time ?

[naive relic]

i was trying to make a tts chatbot

so yeah maybe realtime

[sand iris]

Oh for that purpose you might want to look at https://github.com/w-okada/voice-changer

[naive relic]

what is this?

[sand iris]

real time voice changer

has RVC support

should be straightfortward, there's socketio listening on /test, you send 16KHz audio chunks -> you get converted audio chunks back

Not sure I still have the dumb proof of concept on hand, but (ab)using the app's API to run voice conversion was damn simple

[naive relic]

it has an api?

[rare goblet]

Ayo? @naive relic level 2 !!!

[sand iris]

REST for the settings (it's real messy)

and websocket (via socketio) for inference in real time

[naive relic]

woah

[sand iris]

IIRC there's also a REST endpoint for audio conversion but I haven't checked

It's not documented, it's just there because the app uses it lol

But it should be somewhat straightforward to just open it, go on any web browser, and inspect what it does in the network tab

Throwing another idea, but if you manage sound devices right, you can use the current app directly to convert any audio coming from a specific audio input device

This is getting really messy lol

But it would be something like :

TTS -> Virtual audio cable output -> Virtual audio cable input -> voice-changer voice conversion -> play out the converted audio

[naive relic]

hmm

pretty complex but

you are giving me better idea than GPT lol

[sand iris]

I mean this isn't too far off compared to what people do to voice change in real time

instead of using a microphone as input, you just use the sound coming from the TTS

[oblique lava]

i need help

can someone jump on call with me and walk me through it

[plush pilot]

-colabs

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• GPT-SoVITS (demo), by litagin Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[plush pilot]

-local

[azure marsh]

🖥️ Local stuff

• 🍏 Applio, by IA Hispano GitHub
• Mangio-RVC-Fork, by Mangio621 Huggingface
• RVC Studio, by SayanoAI Huggingface
• AICoverGen, by SociallyIneptWeeb GitHub
• Replay, by Replay Team Website
• Original RVC, by the RVC-Project team GitHub
• GPT-SoVITS, by RVC-Boss GitHub

Need some help? 🤔

You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.

More commands: -audio, -colabs, -overtrain, -realtime, -rvc, -help

[dawn nebula]

@plush pilot pls use #🤖│bots

[dawn nebula]

have u uploaded the vocals?

[tacit grail]

what does increasing the "feature retrievel rate" do in RVC GUI?

[sand iris]

increate the % of "ground truth" where it only picks samples from the dataset the model was trained on

[tacit grail]

i have no idea what you just said

[sand iris]

audio model is trained on audio

audio model guesses what it should convert audio to

but audio model can be bad

so instead of only using the audio model, it can use existing audio samples from the speaker (if you're using a spongebob model, it will use a piece of existing spongebob audio)

[tacit grail]

uh what does that have to do with "feature retrievel rate"

im confused

[sand iris]

the rate is a %

at 0, only the audio voice model is used and tries it's best to convert

at 1 only the reference audio is used (from the exact data the model is trained on)

[raven kayak]

guys my one wont work it dosent change my voice

[half cove]

experiment_name: is what it will name your model
pretrain_type: is the type of pretrain it will use personally use ov2 if dataset length under 8 min but if its higher then use original
model_architecture: use v2
target_sample_rate: for most models its 32k if you wanna check put your dataset into a spectrogram and check the cutoff and double it
speaker_id: dont touch it
pitch_extraction_algorithm: the f0 method use i recommend rmvpe
crepe_hop_length: id recommend keeping it the same
pitch_guidance: if you want your model to sing then you need to tick this

[lucid crow]

hello i cant hear myself speaking

can anyone help

[flint vapor]

On okada right?

[lucid crow]

yes

[flint vapor]

Put on monitor your default speakers

[lucid crow]

hi, can u dm me?

tyy

[flint vapor]

Sure

[tacit grail]

how do i add the reverb and back vocals back and make it sound like the ai's voice

[rare goblet]

Ayo? @tacit grail level 6 !!!

[flint vapor]

On okada or where?

[tacit grail]

well im using RVC GUI and UVR and im mixing the sounds using audacity

"mixing" is really just cutting the fat and trimming unneeded parts and make 2 characters sing and take turns and stuff for me rn... idk how to properly mix audio and all that stuff yet

is there a better way of doing this? i like how this method is simple and all but i feel like im heavily restricting myself because i wanna make ai covers actually sound good and not mess up a lot but idk how to "properly" do this

[flint vapor]

Don't use rvc gui because is very old

[tacit grail]

oh?

[flint vapor]

You can make ai covers with the normal inference 😄

But rvc gui is not so updated

and it doesn't have mangio crepe and rmvpe

[tacit grail]

idk what either of those are

[half cove]

f0 methods

[flint vapor]

Are some pitch extraction methods

But the extractions that rvc gui have are outdated

[tacit grail]

ah ok. whats the thing i should be using?

[flint vapor]

The normal inference, you are using local version or colab?

[tacit grail]

id prefer local but whichever gets me best results ill do.. as long as i can figure out how to do it

[flint vapor]

Ok i will send you now a guide 😄

[half cove]

i dont think there is a rvc gui colab so i assume they are using local

[flint vapor]

https://aihubdocs.github.io/en/essentials/how-to-make-ai-cover/ these are all the instruction on how make an ai cover and the tools you need on local or colab

[flint vapor]

Yes true

[finite galleon]

Someone needs to make a new rvc gui fr

[rare goblet]

Ayo? @finite galleon level 32 !!!

[flint vapor]

I miss that tool

[half cove]

-gui

[azure marsh]

https://cdn.discordapp.com/attachments/1122285248844144733/1203460490475343953/caption.gif?ex=65d12cec&is=65beb7ec&hm=bd2fb8d010006dd7c6e3c1c67d3ae846fd1478e1a3124c544c31b43086fe54aa&

[flint vapor]

i used in june

but was having always problems

[flint vapor]

If i'm not wrong someone already did it or was working for do a new version

[finite galleon]

Good then

(even tho i only use colab)

[flint vapor]

I have rvc gui in my old hard disk, is like 70 gb of stuff inside that folder

I used in the last days of june

[lucid crow]

can someone help me i cant hear myself.

please.

[glad zealot]

start]

[lucid crow]

hindi ko padin marinig

@glad zealot

[glad zealot]

-tensorboard

kewk not that

[rare goblet]

Ayo? @brittle wing level 1 !!!

[glad zealot]

just clean it as much as possible

[latent kettle]

can someone show me an example of over fitting. the graphs mel,Kel . G loss , D loss ... how can I analyze these graphs.

[glad zealot]

cleaning basically means remove anything thats not the voice of the character you want

[lucid crow]

@glad zealot bro dms

[latent kettle]

use tensor board... look at total g and d loss as well as mel, kl

[glad zealot]

bro im not your personal helper

[lucid crow]

uh?

im having problems hearing myself

[latent kettle]

are you using colab ?

[lucid crow]

your goal us to help wtf

[latent kettle]

just below.. where you are training

[glad zealot]

ask in here not in dm

ye lemme find the command

or the site

[lucid crow]

I keep on asking why i cant hear myself

Hindi kanaman sumasagot

[latent kettle]

just here

[proper shale]

give us anything that could help

if you're not willing to do that, im sorry but we cant help you

[latent kettle]

click on reload icon located on top right of tensor board

[glad zealot]

@proper shale you know the tensorboard command or link? cant find it

[lucid crow]

I already sent my settings here.

All i get was ignored.

[half cove]

like the guide or the bat file to open it

[finite galleon]

-overtrain

[azure marsh]

Hey, @glad zealot!

👇 Here are some resources to help you identify if your model is overtraining

How do I know if my model is overtaining?

All-In-One Guide on how to make a good model

This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[proper shale]

Yeah but sadly it turns out that's not the only thing you need to send

There's also the cmd logs, and your specs are slightly important too

[finite galleon]

That command?

[lucid crow]

can you dm me? its more faster there.s

[latent kettle]

can someone show me an example of over fitting. the graphs mel,Kl . G loss , D loss ... how can I analyze these graphs.

[proper shale]

Nah I don't really like DMs

[glad zealot]

images dead but ye thats the one

[proper shale]

Just send screencaps here

[glad zealot]

iirc overfitting is if you put waaay too much dataset?

[lucid crow]

here is my specs.

[latent kettle]

too much? how much..?

[proper shale]

do you have a dedicated GPU

[glad zealot]

like maybe 2 hours + dataset

[proper shale]

overfitting is not that

[glad zealot]

kewk

[proper shale]

it's the same as overtraining xd

[glad zealot]

just different term?

[proper shale]

yeah

[lucid crow]

yes i have

[glad zealot]

aight imma save that in my forgetful brain

[proper shale]

thanks @eager pelican

[latent kettle]

I want how to analyze it with tensor board

[proper shale]

and i do not mean the APU or integrated graphics

[glad zealot]

https://rentry.org/RVC_making-models images died but it should still be useful

[proper shale]

considering you have an AMD processor before the Ryzens I believe you're thinking that's your dedicated GPU, but no that is NOT dedicated, that's integrated

[latent kettle]

I read it many times... but I can't understand.. I need an example... the graphs.. showing overfitting

[proper shale]

unless you have an RX480/580 which is actually pretty based

[lucid crow]

aw.

no one wants to cater me

[glad zealot]

basically if loss/g/total is going higher and higher thats it

[proper shale]

bruh what

look im just trying my best here

what about cmd

[latent kettle]

sometimes it becomes constant.. what does it mean?

[glad zealot]

same the other way, if G going down but D is going up it’s OVERTRAINING```

[glad zealot]

constant?

[onyx ferry]

[latent kettle]

like no change.. no up no down...

[latent kettle]

and what for kl and mel ? are they useless

[proper shale]

kl and mel follow the same principle, if they go up thats a yellow flag

[glad zealot]

i personally just look at g and mel

if they go down gud

[proper shale]

kl = divergence of model compared to dataset, lower better
mel = how good the thing can recreate the clarity of dataset, lower better

[latent kettle]

so everything going down, is good ? ryt ?

[proper shale]

if that happens keep going, if it starts going up thats a yellow flag yea

[proper shale]

yeah

[latent kettle]

ohh .. now I understand.. thanks a lot

[proper shale]

yw :)

[latent kettle]

one more thing.. can you please tell me how much data should I collect for a good model .. like 30 minutes or more than it

[glad zealot]

ye 30 mins is good

[latent kettle]

what about 40 ... or more .. is it bad ...

[glad zealot]

nah still good

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[latent kettle]

so I can consider 30-60 minutes

[glad zealot]

if you want

but 30 is enough

[latent kettle]

ohh .. much batch size for 30 or 30+ do you recommend

[lucid crow]

Should I buy a new mic guys?

[rare goblet]

Ayo? @lucid crow level 6 !!!

[lucid crow]

I can't rlly hear myself on the RVC

[latent kettle]

why ?

[lucid crow]

But my mic works well.

I can't hear myself on the RVC

I tried the monitor and such, I can't hear still. I'm pissed off

[latent kettle]

so you don't need to buy it .. if it is working well

[lucid crow]

But I can't hear myself.

From the RVC

[latent kettle]

try to use headphones for monitoring

[molten pecan]

#🔍│help-w-okada

[latent kettle]

search in #1192011222023950368 or #🔍│help-w-okada

[lucid crow]

I'm using a headphones with mic

I have the okada installed. In having problems with the RVC

I can't hear myself.

[latent kettle]

https://rentry.co/VoiceChangerGuide

[proper shale]

Give screenshots of CMD then

Like... help us if you wanna be helped 😭

[flint vapor]

You selected the default speakers?

In the windows audio settings

See what speakers you have selected from windows settings

[proper shale]

@flint vapor it's a W-Okada problem, not speakers n shit

But eh I guess I can't help much cuz no cmd

[flint vapor]

Oh yes sorry, i just said maybe she/he didn't select the default speakers on monitor and doesn't hear her/himself

[lucid crow]

I can hear my Spotify.

I get errors on cnd

[proper shale]

They do have the speakers on the monitor option

[lucid crow]

But the app still goes run

[flint vapor]

Can we see?

[proper shale]

Alrighty alrighth show them