✨│ai-help

[sand iris]

all the others are to run voice conversion

(so those are the ones you're looking for)

[viral sentinel]

idk if this is the right channel for this but whats the best rvc programto use for the best results or does it make a difference? im kinda new to it

[rare goblet]

Ayo? @viral sentinel level 1 !!!

[junior glade]

-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

[dreamy falcon]

Hi! Is there a way to find the most liked model in "Voice-Models" section

[lilac glacier]

I want to install applio local in mac

[rare goblet]

Ayo? @lilac glacier level 3 !!!

[forest sluice]

Recommend me good ones then?

[forest sluice]

Sadly, none

[dreamy falcon]

sadge

[twin heron]

There's this one by Kalomaze: https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7

[cedar cloud]

Well.... ran aground, again

[rare goblet]

Ayo? @cedar cloud level 2 !!!

[cedar cloud]

Is kalomaze around?

[twin heron]

@acoustic thorn Can you pls fix this error?

[cedar cloud]

Seconding

[brittle wing]

what training?

[lilac glacier]

Cover

[brittle wing]

use Replay

i have it on mac m1. its good

[lilac glacier]

Dm me

[cedar cloud]

The same also happened

[latent kettle]

what replay

can someone tell me the method to install applio on mac

[sand iris]

Are you willing to go through all this only to do voice conversion (not creating new models) ?

If yes, I guess I can help

[latent kettle]

yes.. I just want to make covers

[sand iris]

Alright. Just beware, you'll have to use the terminal for this

[latent kettle]

yes ..

[sand iris]

Just a few things before we start the RVC installer is ass, just making sure it'll work

what's the output of python --version in a terminal, and also the output of brew

[latent kettle]

do you have any guide ? to install it

[sand iris]

Nope

It's supposed to be like one command to run, but the installer is so broken it needs a lot of things that you can't really put in a guide

[latent kettle]

I have Mac studio m2 max 2023 . will it work

[sand iris]

because it varies a lot

[sand iris]

yes

[dreamy falcon]

hi, where do you place the NPY file in RVC models

[rare goblet]

Ayo? @dreamy falcon level 1 !!!

[sand iris]

if I remember correctly that's for SVC models

[dreamy falcon]

thanks

[latent kettle]

do mangio rvc works for mac

[sand iris]

Yes but it's bad

the original RVC works much better

and I'm not sure about applio, last time I checked they didn't really test on mac

[latent kettle]

I just want to make covers ..... no matter its applio , Mangio or Original RVC

[sand iris]

good

[sand iris]

need to know this first

[latent kettle]

and I don't want to use colab 😶‍🌫️

[latent kettle]

yes .. can you wait for some time ? I'm not at my home. nearly 30 minutes

[sand iris]

Oh sure

Sorry I thought you were ready

[latent kettle]

you don't need to say sorry.. that's my fault

[brittle wing]

GahDur is bacckkkk

[sand iris]

[brittle wing]

i am rn training my model on RVC disconnected

@sand iris

[sand iris]

kkkkek

I wonder if training could work on mac to be frank, but the code is so damn messy to nagivate

[brittle wing]

no hopes

from my side

[low shard]

yea i dont think so

[sand iris]

It's just a matter of the metal backend supporting the operations RVC needs ¯_(ツ)_/¯

[brittle wing]

should i try this?

@sand iris

[sand iris]

no !

mangio's mac installer is straight ass

garbage

👎

[brittle wing]

ok

[sand iris]

The "original" rvc project has worked for other mac users

[brittle wing]

O_O

[sand iris]

...with some tweaking required depending on the system

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI

[brittle wing]

u know the tweaking?

[sand iris]

direct dl link : https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/archive/refs/heads/main.zip

[sand iris]

It doesn't work for training still

[brittle wing]

bruh.

[sand iris]

But it's mostly the same stuff you did to have it work

[brittle wing]

i could use replay

[brittle wing]

NO solution for local training on mac?

[sand iris]

Not that I'm aware of

at best I could try to get a MacOS VM to do some testing... but that's a pain

[brittle wing]

how will i know when it will stop?

[sand iris]

not sure about that chief

[dreamy falcon]

is there an exact limit on how long the audio for RVC can be? (I tried an hour long audio - failed)

[proper shale]

not really, but honestly I would not recommend using anything above 40 mins

[dreamy falcon]

is it quality? (when going above 40)

I mean, does it get worse? Or why you wouldn't recommend going past 40min

[ruby wedge]

how do i make the voice less robotic?

[rare goblet]

Ayo? @ruby wedge level 1 !!!

[proper shale]

because it doesn't get any better

[dreamy falcon]

understood

thanks!

[proper shale]

yw :)

[proper shale]

if using W-Okada: https://rentry.co/W-Okada-FAQ

[ruby wedge]

whats the difference between okada and rvc?

[proper shale]

okada uses rvc for realtime conversion

normal rvc does audio file conversion

[ruby wedge]

ohh

for rvc then

because some parts sound really good

but some others are really robotic

how do i make it constant?

[dreamy falcon]

(after some testing found the cap for RVC audio length to be about 10min; maybe 15min) - GTX16050ti

[brittle wing]

75 epochs till now 😦 very slow

[proper shale]

ig try getting better audio or something. normally it's the audio's fault

or maybe the mode

[latent kettle]

when I used this command it says command not found

[sand iris]

both ?

or only python ?

[proper shale]

oh that explains it

4gb VRAM won't cut it for local training I ain't gonna lie, you'll have very slow training sessions

[latent kettle]

yea only python

[proper shale]

(cause you need to use batch size 2)

[sand iris]

and brew displayed something ?

[latent kettle]

it also says command not found

[sand iris]

For python that's good

for brew though

@proper shale do you have brew on your mac / know how to install it properly ?

[proper shale]

don't have it, don't know how to install it properly

[latent kettle]

no .. I don't know and I don't have it

[proper shale]

😭

[sand iris]

apparently

https://brew.sh/

[ruby wedge]

how should i do the settings?

[sand iris]

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

this unholy command does it

[proper shale]

mess with pitch options n stuff, see what works best

[ruby wedge]

what does crepe hop and feature retrieval rate do?

[latent kettle]

what is this

[sand iris]

installation command for brew, according to their documentation

[sand iris]

that's the funny macos stuff

which is part of the reason why getting RVC on macos doesn't have any proper guides

[latent kettle]

python

[small burrow]

what happened to mizuki akiyamas rvc did it get deleted 😭

[sand iris]

oh wait python is installed ? Ah

[latent kettle]

yeh

[sand iris]

...That's even more fun now !

because RVC needs python 3.8, 3.9 or 3.10

other versions just do not work !

[small burrow]

on google colab whatever rvc i input after the install it doesnt work

[latent kettle]

so I have to install python 3.9 or 3.10

[sand iris]

yes

and also modify the RVC Installer to use that version instead of the default system one

[small burrow]

help some1????

[latent kettle]

okay .. after I'm going to install 3.10.11

[sand iris]

what do you mean ?

[small burrow]

the google colab rvc thing

[latent kettle]

what ...?? how

[small burrow]

web version of rvc

i cant send screenshots bc no perms

[sand iris]

Edit some text, it's simple enough

I could just send the modified script if needed, that isn't an issue

[latent kettle]

what I have to download

[sand iris]

nothing yet

[sand iris]

You see this thing here it's complaining about ?

that xcode stuff is needed later to have RVC run

this whole setup will take a while

gotta go eat, I'll be back

[brittle wing]

who is having trouble on mac?

[sand iris]

Dr. Jr

[latent kettle]

@lilac glacier

[sand iris]

they don't have brew installed on their machine, nor the xcode tools

...and python 3.11 is the default version which will cause compat issues

so yeah all the breaking factors are present on their system lmao

[latent kettle]

actually I'm not .. @lilac glacier have this issue.. I'm windows user

[brittle wing]

@lilac glacier to make ai cover, just use Replay!!

[latent kettle]

we manually installed it

[small burrow]

can someone help augh

[rare goblet]

Ayo? @small burrow level 1 !!!

[small burrow]

omfg i just remembered how fucking annoying arcane is

[sand iris]

.

[small burrow]

ok i can send screenies now

[sand iris]

share the link of the colab you are using pls

oh that's the old broken one

refer to

-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]

ilaria rvc is the right one

[small burrow]

whar

ty!

[brittle wing]

whats that

[small burrow]

anyone have a akiyama mizuki model?? 😭

it got taken down

did all pjsk voice models get taken down?

[warm stone]

hi can someone help me coz i tried putting my audio to covert it but i get this error i already put the path but i still get an error

[small burrow]

where do i get these

ty btw!

nvm im dumb

[dreamy falcon]

look website url

[small burrow]

tyyy

[whole lynx]

heyyy guyyy i got this coool idea what wanna prank my friend with and uhhhh can someone help me train? my friend voice

[rare goblet]

Ayo? @whole lynx level 1 !!!

[small burrow]

btw in the web version it showed estimated 2 seconds for the song but its been 5 minutes

[small burrow]

please don't do that to anyone irl that isn't well known or whatever, it's dangerous :3

wh-

[whole lynx]

sooooooooooo

but it will be funny

[small burrow]

just sayinggg your friendships gonna be shattered to pieces if u do the wrong thing

don't use ai voices for malicious purposes

that's kinda illegal yknow

also in the rules if im not mistaken

[whole lynx]

ohh nooo i just want him to do that dream singing the part uhh i cant BREATH when your not there

[whole lynx]

ohhh

whattttttttttt

buttt it not illegal if noone know rightttT?

[small burrow]

still be careful then

[whole lynx]

im jk

haha jk

that was uhh

i meant

it was MY voicee

yep

@small burrow could youuuuu maybe help me do it?

yk

[small burrow]

sorry im just here for song covers :3

[whole lynx]

ohhh

[small burrow]

wanna hear one

i can send in dms

i like vocaloids lol

[whole lynx]

uhhhh

how about no???

[finite galleon]

What prank

[whole lynx]

i justtttt want him to sing the

i cant BREATH when your not there

from dream

[whole lynx]

could you maybe help me outtt?

[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

[finite galleon]

Nah but maybe someone would in #1159289738314919936

[dreamy falcon]

could anyone recommend/send some really good RVC models? I managed to find "bill-wurtz" and "barney-calhoun" models, which are actually very good if used with the right audio. Some that I downloaded seem to have very bad quality (robotic or some other sh*t)

[little island]

If you don't have say 10 minutes of training data for a model, would it be viable to just use say elevenlabs outputs as training data since they can clone voices relatively well even with just a minute of input?

[latent kettle]

can I dm you ?

[latent kettle]

actually I'm occupied now .. I'll install it later!

[flint vapor]

You can train a model with less of 10 minutes dataset on rvc too

[little island]

I guess but would it be better to train a model on say 1 minute of regular speech or 10 minutes of elevenlabs tts assuming the elevenlabs is like 85% of the way there?

[rare goblet]

Ayo? @little island level 2 !!!

[flint vapor]

I recommend rvc because is also free and more professional for cloning voice, but is your choice to where clone

[little island]

the issue with rvc is it doesn't maintain things like accent yeah?

and the tts options are limited

[flint vapor]

You can clone on elevenlabs but you need to pay

but for me rvc is best

[little island]

I do pay for elevenlabs, that's what I said 💀

I think u misunderstood what i asked originally

[flint vapor]

Sorry my bad

What?

[small burrow]

the rvc keeps crashing when i generate the songs aa

ughgtrwanjkef

[little island]

🔥

[small burrow]

anyooone? im not doing anything wrong why does it not generate 😭

nvm it was at rvmpe

[flint vapor]

Try to put mangio crepe

[flint vapor]

Is on colab or hf?

[small burrow]

ilaria rvc

[flint vapor]

No no i know, but you using from google colab or huggingface space?

[small burrow]

huggingface i think?

let me check

yea huggingface

[flint vapor]

You should use on colab because on hf can have problems

i send you the guide wait

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

[small burrow]

now its saying error

[flint vapor]

Yes on hf you can encounter some error

[small burrow]

i refreshed like 6 times already...

[flint vapor]

Don't use on huggingface now

[small burrow]

but its comfier aaa

whatever i guess

gradio or cloudfare?

[flint vapor]

Gradio

Follow the guide

[small burrow]

whys it taking os long aaa

[flint vapor]

Don't worry is normal

[small burrow]

who wanna hear the cover after i make it :3

is it permanent btw

wait

[flint vapor]

I think they removed the channel for upload cover, you can only upload on yt and after show to someone

[flint vapor]

You installed before?

[small burrow]

yeah???

installed it then opened it

[flint vapor]

Try to refresh and open again

If will not work try this https://aihubdocs.github.io/en/rvc/cloud/aicovergen-ui/

[small burrow]

same error

[flint vapor]

try this

[small burrow]

ima try cloudflared

ok

[flint vapor]

You have to select gradio

[flint vapor]

Do what is written in guide

[small burrow]

what am i supposed to do i cant even access the ui

[brittle wing]

ayy my rvc-disconnected still training 🎉

[brittle wing]

on mac?

[small burrow]

windows

nvm got it to work

[rare goblet]

Ayo? @small burrow level 3 !!!

[small burrow]

when did i get to level 3 what...

[small burrow]

how long is it supposeed to take on average to download btw

it keeps redownloading the same shit

[flint vapor]

First time is always long don't worry

[small burrow]

it speeds up overtime?

now it just pause

d

[rare goblet]

Vote Booster: Vote now for a 10% boost. https://arcane.bot/vote
✨ Tip: Use /card to modify your rank card

[small burrow]

whys it uninstalling plugins

jesus fuck

bro is doing a quintuple check

[brittle wing]

what u installing?

[small burrow]

FINALLYYY

FINISHED

[small burrow]

dark souls remastered

jk

just rvc

so i can make a silly kamippoina cover and show you guys

[flint vapor]

Windows office 2000 crack

[small burrow]

nah minecraft infinite diamond hack glitch 100% free 2024

[left oasis]

why?

[flint vapor]

Super official crack

[flint vapor]

Ilaria rvc is broken for now i think

https://aihubdocs.github.io/en/rvc/cloud/aicovergen-ui/ follow this for now

[small burrow]

who wanna hear the cover after its done

i gotta edit the bg music tho

[flint vapor]

You can send by yt link

[small burrow]

dms?

its SUPER good

[flint vapor]

Ok 😄

[small burrow]

gimme a minute

[whole lynx]

ohhh alr thanks

[rare goblet]

Ayo? @whole lynx level 2 !!!

[fading summit]

-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]

sure

[rotund badge]

Hi everyone, how can I use new RVC?

[left oasis]

thanks

[flint vapor]

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

[flint vapor]

You welcome 😄

[rotund badge]

THANKS

[flint vapor]

You welcome 😄

[night relic]

Does anybody know how to download m3u8 files on pc?

Mention website or software name.

[molten pecan]

@flint vapor reminder that the Ilaria RVC guide has a troubleshooting section

Same with Mainline, Mangio & Applio

And UVR

[flint vapor]

Ok

[molten pecan]

Any suggestions for the docs you lmk =3

[woeful sedge]

this sucks

ill try another colab

[molten pecan]

@glad zealot Hina pls fix

It's been happening for some people

[woeful sedge]

i think the problem is on the install ilaria rvc part

[molten pecan]

Really? Why?

[woeful sedge]

now that i sound it out, it makes even less sense

sorry

[molten pecan]

Apology not taken !!

:3

[glad zealot]

reeeee

[woeful sedge]

: (

[glad zealot]

just fixed that earlier

[woeful sedge]

specify earlier

[molten pecan]

Hina do you sleep?

[glad zealot]

around maybe 4 hours ago?

[glad zealot]

yes

maybe..

Shrug

[forest sluice]

Make a new cell, then type in !pip install ffmpy?

[woeful sedge]

im using applio rn soo

[rare goblet]

Ayo? @woeful sedge level 2 !!!

[brittle wing]

Does anyone know how to separate the lead vocal from the backing vocals?

[molten pecan]

With UVR

Using the BVE model

[woeful sedge]

i personally use a website

[molten pecan]

https://aihubdocs.github.io/en/ check the Vocal Isolation page

[brittle wing]

It's too heavy for my PC

D':

[molten pecan]

You can use it on the cloud though

With Google Colab

=3

[glad zealot]

no

[brittle wing]

Do you have the colab link?

[glad zealot]

disconnect and retry run everything it should work now

[molten pecan]

It's in the docs

Along with a usage guide

[woeful sedge]

alright thanks

[brittle wing]

It's good to use Colab, the problem with it is that it disconnects by itself, especially in the middle of training, this happened to me the other day.

[glad zealot]

you need to contantly keep checking the colab tab

also remember that on free tier its only 4 hours a daw

day

[brittle wing]

I know, it's sad hdahsdhsa

[woeful sedge]

and dont forget the fact that it doesnt allow webuis

[glad zealot]

yesnt

Ummm

[woeful sedge]

[night relic]

Kindly reply

[flint vapor]

We don't know where, this support chat is about rvc

[night relic]

What's rvc?

[sand iris]

like the file itself ? Or the stream contents behind it ?

ffmpeg is more often than not always the solution when it comes to multimedia, but it isn't really convenient as it's Command Line only

[forest sluice]

Look for m3u8 downloader extensions on your browser

[night relic]

Video file and the audio

[night relic]

Link please

[glad zealot]

google

[forest sluice]

FetchV video downloader (for Chrome) is what I use

[dark sun]

yo, what do i do if i have that infinite "wait web server..." thing?

[sand iris]

scroll up and screenshot the lines above the "Wait web server" spam

[dark sun]

sec

where do i send it? in dms?

the pic

[rare goblet]

Ayo? @dark sun level 1 !!!

[night relic]

I tried it but the video link does not have m3u8 extension on it, would you try downloading it, to see if it works on your side?

[dark sun]

i sent pic in dms

ok

[sand iris]

try removing she stored_settings file

and relaunching

might work

[mighty fog]

hey i'm trying to use the llaria rcv file

but idkhow to put the models

how do i do it ?

[sand iris]

at the top of the thing, there is another tab right ?

if you put a link to a model, it should work

[sand iris]

links that you can find in #1175430844685484042

[mighty fog]

okay !!!

and where should i put the links

i don't see where ican insert them 😭

[sand iris]

you see the two "launch" buttons ?

you have to execute the install one

and once it is done, the last one

hang on

[simple bone]

guys i keep getting this error JSONDecodeError: Expecting value: line 1 column 1 (char 0)

[sand iris]

-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]

Use Ilaria RVC linked above instead

[simple bone]

yes thanks it seems i was using an old version

[rare goblet]

Ayo? @simple bone level 1 !!!

[mighty fog]

i executed

i am here

where do i link the models ?

[sand iris]

oh perfect

see the thing at the top named "download voice models" ?

right there

[molten pecan]

https://aihubdocs.github.io/en/ check the Ilaria RVC page for a usage guide =3

[mighty fog]

ohhhh

[sand iris]

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

@molten pecan automaze still links to the old docs on the aihub.wtf domain

finding the link to the new ones was a pain lol

[molten pecan]

Yeah, I am working on revamping the commands but

Tea told me to stop and not touch them until I hear back from her

[rare goblet]

Ayo? @mighty fog level 2 !!!

[mighty fog]

i put the exact link and it still doesn't work 😭

[dark sun]

Still doesnt work ok

Still the wait web server thingy

I will just

Delete and download again

[misty forge]

Should I press this?

I have a three minute dataset, for reference.

[flint vapor]

I always put this to no

but if you are on colab

you can put also yes

[misty forge]

I'm on local with a 3090...

So I just don't know.

[flint vapor]

In this case i always put no when i train

i train on local too

[misty forge]

What do you do for the batch size too?

[flint vapor]

depends from the dataset you use, how long is?

[misty forge]

3 minutes and six seconds.

[flint vapor]

There is not a precise one, is always about dataset, sometimes i use 8 or 10

[flint vapor]

If is small you can also put yes

[flint vapor]

here says that is ideal for small datasets

but i always put this as no

[misty forge]

Okay... by the way, I'm using RIN_E3 as the pretrain. Should I switch it to something else?

[flint vapor]

For small dataset i recommend Ov2

RIN is for long dataset

[misty forge]

These, right?

[flint vapor]

Yes

[flint vapor]

Before select 40 are you sure you put 40k?

[misty forge]

Didn't touch it.

[latent kettle]

hey.. can you tell me about these pre trainers ? what is Ov2 and RIN_E3

[misty forge]

What about the epoch amount? Right now I have the save frequency for 50 and the total epochs set to 300. Should the total amount be less or more?

[flint vapor]

Have you put v2?

[misty forge]

Yes, I did.

[flint vapor]

Ohh why you don't have also 32k version?

[flint vapor]

Epochs for small dataset with ov2 you can put also 100

or 80

[flint vapor]

Which version of rvc you using? Mainline, mangio or applio?

[half cove]

for 32k

click v1 then v2

to load it

[flint vapor]

Yes try to refresh

[misty forge]

I don't think I'd ever use 32k, though...

[half cove]

not refresh

[rare goblet]

Ayo? @misty forge level 2 !!!

[half cove]

click v1 then vv2

[half cove]

most models should be 32k

[flint vapor]

No like to refresh the bar clicking v1 and after v2 sorry my bad😭

[misty forge]

It worked!

[flint vapor]

Fine! 😄

you can download the 32k fix for ov2

Wait, which version you have of rvc?

Applio,mangio ? Mainline?

[half cove]

@flint vapor i saw a ping did you have any problems?

[misty forge]

1006 V2 of mainline.

[flint vapor]

Yes i had a problem with ov2 but i resolved 😄

[half cove]

what was the problem

[flint vapor]

When i was training with ov2 the training was stopping after 1 minute, i close and opened again rvc interface and was work fine

Was like a general problem

But was all ok after 😄

[flint vapor]

I recommend you to use applio or mangio

they have more features

[misty forge]

Like...?

[flint vapor]

Wait i send you a doc, so you can see all 3

and choose

[half cove]

so after restarting it fixed?

[flint vapor]

https://aihubdocs.github.io/en/ Go on RVC

[flint vapor]

Local and you can see all 3

[half cove]

shouldnt effect inference anyway

[flint vapor]

Yes but i think i had also the old versions of ov2 32k

because i downloaded again from hf

and they worked

[half cove]

ohh

[flint vapor]

so i think this resolved the problem

and not the restart

[half cove]

so you didnt use the one from me

before

[flint vapor]

but i did both

[half cove]

the torch keys fixed

[flint vapor]

I had the old version from dev updates

[half cove]

the one from me or the main one

both were in dev updates

[flint vapor]

I had the main i think, and after i downloaded yours and it worked 😄

[half cove]

ok nice

just needed to know

[misty forge]

Poopmaster, what should the batch size be for a 3 minutes and 6 second model?

[flint vapor]

Your fix helped me with mangio version

because i had the old one before

so for this didn't worked i think

[flint vapor]

around 6-8 ?

[flint vapor]

idk if also 4

[half cove]

you could try a smaller batch size so more step per epoch

for small datasets ov2 good

[flint vapor]

Ok Nice 😄

[misty forge]

Do I have the right version? of ov2?

[flint vapor]

Yes yes

[latent kettle]

hey.. can someone tell me about these pre trainers ? what is Ov2 and RIN_E3

[flint vapor]

Damn so wait you have 24 gb of vram?

[flint vapor]

Ov2 is generally recommended with small datasets

RIN the opposite

[misty forge]

Yes, I do.

[flint vapor]

Damn

You can do a model of 20 minute of dataset in 30 minutes

[misty forge]

I... was just about to say I got a index out of range error...

[flint vapor]

Oh can i see?

[latent kettle]

why they do this ? as we can also change batch size

[flint vapor]

Because they can help also to improve a bit the quality of vocie

for example sometimes ov2 is more clear

[latent kettle]

okay ...

thanks

[flint vapor]

Oh strange error, you did all the steps first?

[flint vapor]

you welcome 😄

[flint vapor]

Is the folder path of pretrains put good?

[glad zealot]

@half cove howd ya even fixed it?

the pretrain

[flint vapor]

You click train model or one click training?

[misty forge]

train model.

[half cove]

change torch keys

[glad zealot]

ye no idea what that means :D

[misty forge]

These are for pretrained_v2, by the way.

[grim bay]

How to upload the model on Huggingface?

[flint vapor]

You installed them good

[flint vapor]

wait i send you a guide

[flint vapor]

https://aihubdocs.github.io/en/essentials/uploading-models-to-hugging-face/

[misty forge]

Is this config okay?

[flint vapor]

Yes you have the ov2 pretrain here

[sand iris]

pytorch weights are stored in a specific way

2.1 decided to change the attribute names

so they're not accesses the same way as in 2.0

[misty forge]

Uh... I still got the error...

[half cove]

on applio it should fix for you now

so no more problems

[sand iris]

it's backwards compatible, so older saves load fine on newer versions,
but newer saves don't load fine on older versions

[flint vapor]

Have you check if the path where you have pretrain are inside a folder named assets/pretrained_v2?

[half cove]

simplcup just used a version that didnt swap it back

[rare goblet]

Ayo? @misty forge level 3 !!!

[half cove]

it only was out for like a day before being fixed

[flint vapor]

Mmh, can you try change rvc version?

Mainline have less features

idk if mainline have more problems

[half cove]

it works on mainline

[glad zealot]

what is?

[half cove]

@misty forge where did you download it from

[misty forge]

https://huggingface.co/ORVC/Ov2Super/tree/main

[half cove]

thats why then

[glad zealot]

i dont thnk the 32k on that doesnt work

fck grammar

[half cove]

does work just only in applio

[misty forge]

WHAT

[half cove]

use dis #📰│dev-updates message

[glad zealot]

kewk

ye use the fix version

cause its fixed on taht version :D

[flint vapor]

Yes true you need to download the poopmaster fix

[flint vapor]

Because i had problems too with this on mangio

[half cove]

they should just use it on the main one

[half cove]

mangio uses older torch aswel

das why

[glad zealot]

@half cove i sent you the mainline colab right?

[misty forge]

Okay, good, it's working now...

[half cove]

i think

[flint vapor]

Ok 😄

[glad zealot]

what you think i should add on it or change on the colab?

[half cove]

maybe the mangio one

[glad zealot]

so that we can release it

[glad zealot]

same thing

[half cove]

let me check it again rq

[glad zealot]

actually lemme update the mangio one too

[half cove]

it seems good tbh

[glad zealot]

just updated it

[half cove]

but on the pretrains bit use
CustomPretrain="poiqazwsx/Ov2Super32kfix" #@param ["ORVC/RIN_E3","poiqazwsx/Ov2Super32kfix","ORVC/Ov2Super","ORVC/Snowie_RuPretrain_48k"]
pleas

so you dont have to type it

[glad zealot]

lmao

aight

[half cove]

i think i spelt it all right

[rich valley]

I have a problem voice models just doesn't downloading
I installed it saved in cloud
but there still "ModuleNotFoundError"

[glad zealot]

still have to manually puth the path in the ui tho

:D

[half cove]

[half cove]

which colab you usin

[glad zealot]

took me 3 days to make this work

[rich valley]

Easy GUI (for RVC v2, with crepe) (with improved downloader)

[glad zealot]

its good now tho

[half cove]

das old

-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

[glad zealot]

i think thats the outdated version

[half cove]

[rich valley]

thx

[half cove]

ila rvc works now right?

[rich valley]

need check

[half cove]

i think i saw you say that

[glad zealot]

yup

[half cove]

ok nice

[glad zealot]

should

Ummm

[half cove]

[sand iris]

rvc is held together by glue and prayers

[half cove]

indeed

[glad zealot]

agreed

[night silo]

https://prnt.sc/lKAz6bmFFNFL

['infer/modules/train/preprocess.py', 'C:\\Users\\berdoze\\Desktop\\voice', '40000', '6', 'C:\\Users\\berdoze\\Desktop\\Retrieval-based-Voice-Conversion-WebUI-main/logs/mi-test', 'False', '3.7']
C:\Users\berdoze\Desktop\voice/mrbeast.mp3->Traceback (most recent call last):
  File "C:\Users\berdoze\Desktop\Retrieval-based-Voice-Conversion-WebUI-main\infer\lib\audio.py", line 38, in load_audio
    ffmpeg.input(file, threads=0)
  File "C:\Users\berdoze\AppData\Local\Programs\Python\Python39\lib\site-packages\ffmpeg\_run.py", line 313, in run
    process = run_async(
  File "C:\Users\berdoze\AppData\Local\Programs\Python\Python39\lib\site-packages\ffmpeg\_run.py", line 284, in run_async
    return subprocess.Popen(
  File "C:\Users\berdoze\AppData\Local\Programs\Python\Python39\lib\subprocess.py", line 951, in __init__
    self._execute_child(args, executable, preexec_fn, close_fds,
  File "C:\Users\berdoze\AppData\Local\Programs\Python\Python39\lib\subprocess.py", line 1420, in _execute_child
    hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\Users\berdoze\Desktop\Retrieval-based-Voice-Conversion-WebUI-main\infer\modules\train\preprocess.py", line 87, in pipeline
    audio = load_audio(path, self.sr)
  File "C:\Users\berdoze\Desktop\Retrieval-based-Voice-Conversion-WebUI-main\infer\lib\audio.py", line 43, in load_audio
    raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: [WinError 2] The system cannot find the file specified

end preprocess
``` what am i doing wrong ?

[proper shale]

what's the folder path of your dataset

oh yeah nvm its correct

hm maybe try enabling pitch guidance

[young pollen]

Hi guys, Im new at RVC and I and I downloaded voicelines I want to use, I extracted the folder and I dont know what to do next, can someone help me?

[proper shale]

now just read the guides on how to use rvc: https://aihubdocs.github.io/en/

[young pollen]

Wich section?

[proper shale]

depends on what you wanna use

[rare goblet]

Ayo? @brittle wing level 4 !!!

[young pollen]

rvc 2

[proper shale]

yeah but it depends on what fork you wanna use, whether it'll be local or via cloud...

[young pollen]

no clue, I just wanna use the downloaded voices

[rare goblet]

Ayo? @young pollen level 1 !!!

[molten pecan]

Check the AI cover page =3

[proper shale]

you can go back on rvc disconnected, use same training variables, skip preprocessing and feature extraction, train index and export model

[molten pecan]

It's a starting point

[proper shale]

well.. not much you can do in that case

unless you wanna ask the model maker for the dataset

[young pollen]

kk

[young pollen]

Oh I get it now, I have great specs so I wanna use local

[misty forge]

@half cove @flint vapor How's this? This was trained on the settings you guys recommended and I'm not sure if it holds up. 100 Epochs in total... is it good?

[proper shale]

yeah then check the tutorials for local forks

[young pollen]

k

[flint vapor]

Looks ok

[glad zealot]

sounds good to me

[night silo]

yeah already tried alot of things tbh, even tried in a second computer just to be sure

[proper shale]

sounds pretty decent, id say keep training that

[proper shale]

maybe the audio's screwed up... hm

[misty forge]

To how long? It already took a bit for it to reach 100 epochs...

[proper shale]

make ur goal 500, ofc check the graphs - if it does overtrain before that you can stop

[misty forge]

I... use mainline though.

[proper shale]

id try reexporting the mp3 as a .wav via audacity

[proper shale]

...yeah

its

the same?

for basically any fork

[misty forge]

But... I've never seen a tensorboard anywhere...

[proper shale]

you can just resume training

[proper shale]

oooh

you do need to install it yeah

https://aihubdocs.github.io/en/rvc-resources/epochs-overtraining--tensorboard/#installing-opening

tells ya how

[misty forge]

Uh... yeah, I don't exactly know what equates to overtraining...

[proper shale]

it's not a simple task to determine overtraining, so in case you're in doubt you can send the graphs

[odd shale]

Overtraining = If your d/total graph goes up

[molten pecan]

Indefinitely though

It will go up for a short time over the course of the training

[misty forge]

There's... nothing here.

I did what the tensorboard instructions said...

Did I just fuck up?

[proper shale]

in case it actually screwed up just delete the tensorvenv folder and restart the bat - see if any error pops up on cmd

yeah

it does

[misty forge]

...No outward errors but it's still a white screen...

[proper shale]

[proper shale]

hm

maybe check if the localhost link is actually using the 6006 number that the guide shows

...not really?

in fact, ive heard index only brings more artifacts to the result

does it lack the accent tho? eh, slightly - but this is good sometimes

[misty forge]

I don't understand what happened...

Like... is it just incompatible with mainline?

Should I have mentioned that earlier?

I want to use tensorboard, but obviously... I can't. It just shows a blank page...

[small burrow]

I FINALLY MADE SOMETHING WITH THE THING

GOOGLE COLAB

THANKS FOR THE HELP GUYS

i think its rly good for my first time

[flint vapor]

Hi rosa! You can send it by yt link way because is forbidden to upload copyright stuff 😄

[small burrow]

a i see sorry!

[flint vapor]

Don't worry 😄

[small burrow]

its not copyrighted tho

the original song doesnt have copyright

[flint vapor]

Which was the name of the original song?

[small burrow]

kamippoina by pinocchiop

[flint vapor]

I think the instrumental have

We remove copyright because we had a bad experience on october 5 where our server been destroyed

by some copyrighted stuff

So for risk we not allow 😄

[small burrow]

i see

[flint vapor]

But you can send by youtube link 😄

or in dm 😄

[small burrow]

i see

[misty forge]

Kakao, while you're here, can you tell me what's going on with my thing?

[flint vapor]

However the song you send i have it on my psvita theme

[flint vapor]

Sure with tensorboard?

[small burrow]

whats that

[misty forge]

Yes.

[molten pecan]

What do you need help with?

[flint vapor]

You opened tensorboard and is not working?

[molten pecan]

Did you check our docs?

[flint vapor]

Which tensorboard you downloaded?

[misty forge]

I downloaded the bat that was in the guide and put it in the rvc folder to run...

[flint vapor]

Oh the song you said before : Kamippoina by pinocchio, i have it on my psvita theme

this is the psvita

[flint vapor]

Wait i try to see if i have some fix

[misty forge]

But this problem has never been reported in the server before, to my knowledge...

[proper shale]

Maybe you got some firewall or something?

hmmm

[flint vapor]

i have a python file for execute my tensorboard because if not it doesn't work too

i have this file

[grim bay]

What's the best model to extract the vocal for AI cover?

[flint vapor]

You say in the pitch extraction way?

Like rmvpe, mangio crepe?

Or the rvc versions?

[latent kettle]

installing rvc on mac

getting this error.. how to fix it

aria2

[grim bay]

The UVR

[flint vapor]

There are many models, you want isolate from instrumental?

[rare goblet]

Ayo? @flint vapor level 27 !!!

[grim bay]

Yes, I want the vocal to make an AI cover

[flint vapor]

https://aihubdocs.github.io/en/vocal-isolation--datasets/vocal-isolation/

You can see the guides of many models here

and you can see what choose 😄

[grim bay]

Thank you so much

It's been a long time when I use it the last time

[molten pecan]

For a starting point check the AI Cover page on the docs =3

[flint vapor]

You welcome 😄

[misty forge]

This was the original run with 100 epochs... how is this?

Like... what am I supposed to look out for again?

[rare goblet]

Ayo? @misty forge level 4 !!!

[misty forge]

Seems like it got to its lowest on step 800...?

[molten pecan]

https://aihubdocs.github.io/en/ check the TensorBoard page

There's a usage guide

=3

[misty forge]

So it's when it reaches the bottommost and then starts going up, right?

Because this is where it reached the bottommost.

[molten pecan]

That's not the graph you are supposed to look after

(unless you are an advanced user)

Like I said, check the docs

You'll find a tutorial for newbies

[misty forge]

This one!

The guide says for this one!

Oh...

Oh.

That's massively overtrained...

[molten pecan]

Uhhh what's the smoothing value?

[misty forge]

This is the lowest it went.

[misty forge]

What do you think?

[molten pecan]

Look at the left panel

There's a Smoothing slider

[molten pecan]

It's supposed to be in 0.987

Pls read the docs thoroughly

[flint vapor]

Have you fix with the file i give ? 😄

[molten pecan]

For how long have you been training?

[molten pecan]

This doesn't look good. Seems like TB is bugged

Which batch size did you choose?

[young halo]

Is there any way for the pth files to save automatically on drive? It's like the 3rd time my training interrupts and it won't even reach 200 Epochs before stopping (my model was already trained 120 epochs and I'm trying to resume)

[molten pecan]

And how long is your dataset?

[misty forge]

Three minutes and 6 seconds.

[molten pecan]

Are you using RVC Disconnected?

[molten pecan]

When TB opened, did the graphs take a long time to react?

[young halo]

Yup

[misty forge]

I... don't know? When I refreshed the page, it updated the graphs...

[molten pecan]

https://aihubdocs.github.io/en/ check the RVC Disconnected page. I included how to retain resuming from where you left

[rustic oar]

Traceback (most recent call last):
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/routes.py", line 321, in run_predict
output = await app.blocks.process_api(
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/blocks.py", line 1007, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/blocks.py", line 953, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/components.py", line 2076, in postprocess
processing_utils.audio_to_file(sample_rate, data, file.name)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/processing_utils.py", line 206, in audio_to_file
data = convert_to_16_bit_wav(data)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/processing_utils.py", line 219, in convert_to_16_bit_wav
if data.dtype in [np.float64, np.float32, np.float16]:
AttributeError: 'NoneType' object has no attribute 'dtype'

I'm trying to make a cover

[molten pecan]

And which PTH you should download

[molten pecan]

I think TB is bugged. It's a problem Colab's been having lately

[misty forge]

Why say that?

[molten pecan]

@half cove can you take care of that?

[rustic oar]

help pls

[molten pecan]

Colab's been having problems lately

Weird updates

[half cove]

what r u usin?

[misty forge]

No, I mean like... what gave it away that TB was bugged from the graph?

Is it possible it's just grossly overtrained?

[molten pecan]

It looks too "straight & sharp"

[rustic oar]

i use RVC-beta0717

[molten pecan]

A normal graph would have more smaller variations

[young halo]

I was about to save the 200 Epochs pth file but it interrupted a around 180 and I wasn't able to save it 🥲

[misty forge]

But... wouldn't that mean they just updated the look to be more straight and sharp?

[half cove]

thats old btw also what was the path of the audio you used?

[molten pecan]

Uhh nope. TB is its own independent thing, Colab aren't the developers or anything

[rustic oar]

the path to the audio is correct. I've been checking

[molten pecan]

I had that same problem too

[half cove]

show me

[molten pecan]

Are you using TensorBoard?

[misty forge]

How did you fix it?

[molten pecan]

Or inputting a semi-arbitrary amount of epochs?

[young halo]

Yes, I always do!

[molten pecan]

Oh great then

[molten pecan]

Read the docs. I taught how to retrain and save the PTH

[molten pecan]

Colab needs to get fixed first

[rustic oar]

what

[molten pecan]

Or the person behind the Colab space

[half cove]

show me your file path

[misty forge]

So, you're basically saying I can't use TB right now until it's properly fixed?

[half cove]

make sure nothing has any spaces

[rustic oar]

everything was working fine a week ago

[half cove]

like "ai cover"

[misty forge]

How else am I supposed to check that a model isn't overtrained then?

[flint vapor]

the space between ai and cover

can do some problems

rename the folder aicover

[rustic oar]

oh

[molten pecan]

Well you can, it's just that it's bugged and you can't make proper use of it

[rustic oar]

i try

error

[molten pecan]

Colab is weird like it. It'll get fixed sooner or later

[rustic oar]

Traceback (most recent call last):
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/routes.py", line 321, in run_predict
output = await app.blocks.process_api(
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/blocks.py", line 1007, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/blocks.py", line 953, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/components.py", line 2076, in postprocess
processing_utils.audio_to_file(sample_rate, data, file.name)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/processing_utils.py", line 206, in audio_to_file
data = convert_to_16_bit_wav(data)
File "/home/.pyenv/versions/3.10.11/envs/rvc/lib/python3.10/site-packages/gradio/processing_utils.py", line 219, in convert_to_16_bit_wav
if data.dtype in [np.float64, np.float32, np.float16]:
AttributeError: 'NoneType' object has no attribute 'dtype'
2024-02-26 20:17:09 | INFO | httpx | HTTP Request: POST http://localhost:7865/api/predict "HTTP/1.1 500 Internal Server Error"
2024-02-26 20:17:09 | INFO | httpx | HTTP Request: POST http://localhost:7865/reset "HTTP/1.1 200 OK"

[misty forge]

Again, I need to ask... how else am I supposed to check that a model isn't overtrained, if TB is bugged?

[molten pecan]

@half cove can you double check if TB is working in Colab?

[molten pecan]

If you have good specs, use RVC locally. If you have 8 bucks, use Paperspace

[misty forge]

Uh... one problem.

I have been using all of this locally.

[molten pecan]

OH WHAT

[flint vapor]

He is in local with a 3090

[young halo]

Is there any guide for the cloud version of applio?

[molten pecan]

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

=3

[half cove]

anyone have any problems?

[misty forge]

I do...

[molten pecan]

@misty forge

[misty forge]

Apparently my TB is bugged.

[molten pecan]

Their TB looks really off

[young halo]

Thank you so much! I also might try using it locally, hopefully it doesn't gives me a lot of problems like RVC

[molten pecan]

Well, Applio is RVC hehe

[half cove]

graph could have just fucked up ig

what pretrain they usin

[molten pecan]

Applio is a fork of RVC

[misty forge]

Ov2.

[young halo]

Ohhh, lol, maybe locally could work better then

[half cove]

ov2 does seem to have goofy graphs sometimes

[flint vapor]

If you have nice components yes 😄

[rustic oar]

help

[half cove]

just try 500 and check a few different epoch checkpoints

[rustic oar]

I will try to reinstall the libraries

[misty forge]

Saving frequency to how much? 20?

[half cove]

should be fine

[young halo]

Which components do I need? 👀

[flint vapor]

Especially gpu

but also a good cpu, ram

but gpu is the most important

[misty forge]

And batch size?

Last time I had it on 4...

[half cove]

usually use 8

i recommend testing out different batch sizes after getting used to training models

[latent kettle]

rvc on mac .. getting error

[proper shale]

maybe ya didnt select something properly

[rustic oar]

this is error of lib

[sand iris]

I think this is the random error, but not sure why it even happens

[young halo]

Is that folder created on applio or file Explorer of the pc?

[flint vapor]

Is the main folder of applio in file explorer of the pc

[young halo]

I have it on desktop, idk if that's what u mean 😅

[flint vapor]

You don't have the applio folder? what you have on desktop?

[polar plaza]

Um??

[young halo]

I'm extracting the zip file

[rare goblet]

Ayo? @young halo level 10 !!!

[flint vapor]

Fine

Follow all the guide

[polar plaza]

Why is Ilaria broken too😭

[flint vapor]

https://aihubdocs.github.io/en/rvc/local/applio/

[young halo]

Ok! So that folder with the model's name should be inside that Applio's folder right?

[flint vapor]

Yes 😄

i recommend you to follow all guide because i don't use applio

so i'm afraid to wrong some step

[young halo]

Tysm!

[flint vapor]

the guide is more expert than me 😄

[flint vapor]

You welcome 😄

[young halo]

Oh, so then applio > logs > model folder?

[misty forge]

So these are at: 20, 40, 60, 80, 100, and 120 epochs respecitvely... I have saved all of them. Which one should I go with, and should I train further?

[rare goblet]

Ayo? @misty forge level 5 !!!

[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

[young halo]

I'm training a model on applio but the tensorboard just doesn't show anything at all...

[lilac glacier]

Hii

RVC webui in mac studio
Stucking on processing
Anyone help me ?

[dusty apex]

hi there. I have an AMD GPU, specifically 7900xtx nitro+ and for some reason the voice changer keeps using only the CPU even though I set gpu1. Is there any way to fix this? I use the directML version.

[sand iris]

unless you have integrated graphics, set to gpu0

[dusty apex]

Just to be sure I tried all gpu options and it still uses only cpu

[sand iris]

you're using a custom model ?

you also need to export to onnx, reimport the converted model & use it

[young halo]

Yeah, I did that already but it's not working

The training still going but it's not showing anything on Tensorboard

[lilac glacier]

Plz help

[tranquil cliff]

are you using applio right?

I know your problem

you pressed Start Training before Generate Index

[charred topaz]

Im having the exact same problem I had with WEB UI, in WEB UI, the second and after voice models i created create epochs toooo fast and the result is really bad. I switched to Mangio, and the first voice in Mangio was really good, but then the second voice i trained loads epochs toooo fast

[tranquil cliff]

mangio crepe?

[fiery basin]

Hi, I need some help. I did an RVC training but accidentally deleted the .pth file. I still have the untuched folder in logs. Is there a way I can quickly make a .pth file with the logs folder? I used Mangio-RVC

[charred topaz]

like they create epochs every seconds, and my first voice models i trained in web ui and mangio took their time, like 2 mins per epoch

[charred topaz]

crepe?

[tranquil cliff]

no, I didn't do anything wrong

[flint vapor]

You don't have a backup?

If you don't have a backup i think you have to start all again

[tranquil cliff]

kakao

[flint vapor]

but wait

in logs you have

the audio files

already processed

[fiery basin]

yes

[flint vapor]

Hi ange 😄

[dusty apex]

Looks like I'm really stupid. I found out that through gpu0 it runs normally. I could have sworn it didn't work before. Thank you very much❤️

[tranquil cliff]

ma c'è un programma che si chiama magio propio?

[flint vapor]

No no, mangio è solo una versione di rvc tipo applio 😄

[flint vapor]

But you have to generate the pth again

so i think you have to start all from 0

😢

[fiery basin]

😦

[flint vapor]

put same settings again

and restart the train

so you don't lost the logs file

[tranquil cliff]

ahhh

c'è propio un interfaccia

?

se si mi invii il link?

[fiery basin]

But the logs folder is 2 GB, can't I use any of that info?

[rare goblet]

Ayo? @fiery basin level 1 !!!

[flint vapor]

Si è diversa 😁

Fra poco ti mando il link 😁

[fiery basin]

Also, I have the D and G .pth files, just not the final one

[flint vapor]

Try to put 10 epochs for test and see if generate you a normal pth

Like try to train another 10 epochs

See if you will have a good pth or bad

[flint vapor]

https://github.com/Mangio621/Mangio-RVC-Fork