✨│ai-help

[covert sedge]

oh damn

[inner hazel]

"NFO:mim:====> Epoch: 1 [2023-12-26 21:59:59] | (0:01:36.458898)"

[violet heron]

This is false

It really depends what’s in your dataset and quality of it as well

[covert sedge]

ah

it's in pinned messages and i used to follow this during my model creations

[violet heron]

Also imo your wasting epochs if your dataset if very long

[covert sedge]

i used 350 on 17min dataset and it's not bad

only problem is a high notes but i can prevent them with rmvpe method

[inner hazel]

i lowered mine about 65% now

[calm crown]

Does anyone has an applio colab link that works?

[violet heron]

That’s perfectly fine

[violet heron]

No

[covert sedge]

also is it possible to train an already existing model? i know that mine isn't overtrained yet but i already used google colab limit hours ago

[violet heron]

Train a model that’s already trained?

Or resume training?

[covert sedge]

like i've already made a model right

like a 350 epochs model, and is there a method to add, let's say 50 more epochs to it instead of creating whole new 400 epochs model

[weak elm]

how do you use rvc on discord?

[dusty basin]

Anyone knows how weights.gg achieves a stereo output with rvc?

[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

[proper shale]

if you do the retraining steps and set total epochs to 400, u can do it

keep in mind you need to keep the d and g pth files

[proper shale]

what rvc

[weak elm]

i figured it out

[proper shale]

probably just making the mono output stereo, but it still outputs the same thing in both sides probably

[covert sedge]

i have g and d pth files on my google drive but where are these retraining steps?

[proper shale]

What Colab are you using

[covert sedge]

Disconnected

[proper shale]

Yeah there are steps marked with (for resuming)

Use those and the training step for retraining

[covert sedge]

i closed the page hours ago so it probably won't work

[inner hazel]

damn how long does the index training take?

im using rvc disconnected

"Minibatch step 64/162512"

like how does it take 162512 minibatches

thats like 5h

[astral perch]

Has the mangio rvc fork become the main model trainer now?

[proper shale]

it has always been that

u can use the original RVC too

[astral perch]

The rvc that i used always only had harvest dio and pm as pitch extractions

[proper shale]

oh yeah definitely consider upgrading

[astral perch]

i am installing applio now

[proper shale]

good luck, it's a very problematic install but if it works it's great

[astral perch]

why is it problematic?

[molten pecan]

considered complicated to install by many

[molten pecan]

uhh did you use RVC-GUI?

[sharp island]

HOW

HOW RHE FUCM DID THIS HAPPEN

@pliant crescent HELP

[pliant crescent]

I need more context

[sharp island]

The the gpu ram keeps going up to 15.0 when training

[pliant crescent]

How big is your dataset and please don't make your batch size more than 20

[sharp island]

It is more than 20 minutes •_•

[hushed fjord]

Andreea Esca ?

Who got the Voice

[pliant crescent]

It's unlikely it exists, try finding it in #🔍│find-models

[pliant crescent]

Bruh, that's too much

[sharp island]

Is it affecting the system ram too

Cuz that rises up too

[sharp island]

Wait shit I made the batch 40 it’s my dataset that’s 22 minutes

[pliant crescent]

BRUH

[sharp island]

So is making the dataset shorter the only fix I should do?

[pliant crescent]

Just do 16

[pliant crescent]

Batch size

[proud elbow]

it's fine as quality matters, but the recommended batch size is 8 or 16

[pliant crescent]

In settings

[sharp island]

What about Epoches?

[pliant crescent]

500+

[sharp island]

Any specific number is just at 500?

[pliant crescent]

Are you going to check for overtraining or just trust your gut

[sharp island]

I dunno

Just looking for one more thing I should do to prevent messing up

[pliant crescent]

Just do 500

[soft palm]

i'm trying to use the "CoverGen_NO_UI_v2_en" colab but when I do the mount drive it gives me the "Error: credential propagation was unsuccessful", anyone know a way to fix it?

[sharp island]

Am I overtraining yet?

[pliant crescent]

Go to loss/g/total graph

Use the search bar

[sharp island]

Alr here’s what I got

[proud elbow]

uncheck "ignore outliers"

[pliant crescent]

Okay now max your smoothness

[sharp island]

With the scaler?

[pliant crescent]

Okay good now expand your graph

[sharp island]

How do I do that

[pliant crescent]

Square thing

Next to pin

[sharp island]

Enable range? I’m not seeing a pin

[proud elbow]

that's why I said to uncheck "ignore outliers" if you don't want the graph look cut out

[sharp island]

Ok I did that

[sharp island]

Ok I clicked this thing

[sleek nova]

Moved from RVC main to Applio and now every epoch is exactly 1 step and the models do not train. Interesting.

[pliant crescent]

Yes that makes it bigger

[sharp island]

Here a better photo

[pliant crescent]

Give it more space

[pliant crescent]

^

[sharp island]

How would I do that

[pliant crescent]

Can't you like use your mouse to change the size of the ui

[sturdy palm]

Hello friends, I know the sound models, but I don't know where the colab link is, can you briefly summarize it?

[sharp island]

I don’t know how

Also this is gone

[pliant crescent]

https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp=sharing

[sturdy palm]

Thanks very much

[pliant crescent]

Click the triangle thing

With the inward arrows

[sharp island]

I clicked that the one next to the pin they’re still not showing

[rare goblet]

Ayo? @sharp island level 7 !!!

[sharp island]

It happened when I did this

[pliant crescent]

@proud elbow help me here

I told you click the button left to the pin

[sharp island]

Oh ok it works

They’re back

[pliant crescent]

See

[sharp island]

Anything else I should do?

[pliant crescent]

Your graph needs to be wider

[sharp island]

Ok how do I do that

[proud elbow]

probably no more idea cause I have abit different tensorboard layout from his

[pliant crescent]

Can you hover your mouse on the edge of the right UI

[sharp island]

The graph?

[pliant crescent]

And move the UI to the right

[sleek nova]

I had to reinstall every dependency for Applio to stop training 1 step per epoch and taking .14 seconds per epoch to complete..

[proven hill]

MERRY CHRISTMAS AI HUB!

Celebrate Christmas with the highest quality girl voices for the voice changer or regular rvc v2!

Soft Speaking Female: https://ko-fi.com/s/9c8a5b192e

American E-Girl: https://ko-fi.com/s/e21ec90d1b

Cute & Calm Girl: https://ko-fi.com/s/c8cf0c2129

And you will find many others here, FOR FREE!! https://ko-fi.com/ilariaowo/shop

Have a great Xmas Eve!

[pliant crescent]

Lmao

[sharp island]

I think I did it right

[pliant crescent]

Can you tap Ctrl then -

[sharp island]

Ooooohhh

[pliant crescent]

BRUH

Your gui was zoomed or your monitor is small

[sharp island]

I didn’t know what you meant

[pliant crescent]

It's okay

[sharp island]

Now what?

[pliant crescent]

If the line starts rising, it's overtraining

[elder coral]

what is error make sure make sure your data was preprocessed mean in feuture extration is easygui colab

[sharp island]

What do I do then?

[pliant crescent]

You can stop training, make sure you turned on auto save

[sharp island]

Is there a way to tell if it’s on?

[pliant crescent]

And you've made your index already

[pliant crescent]

It's says in the settings

[sharp island]

Are these the one

[pliant crescent]

RVC settings

[sharp island]

Where’s that at

[pliant crescent]

The part where you enter numbers and check things

[sharp island]

Yeah I’m extremely unfamiliar with RVC can u be more specific

[pliant crescent]

Where's the part you entered your epoch number

[sharp island]

Is this it

[pliant crescent]

Absolutely not

[sleek nova]

Getting mileage out of that helper role.

[pliant crescent]

Could you tell me which colab you're using

[sleek nova]

Looked like it said RVC 2 Disconnected.

[sharp island]

RVC V2

Disconnect

[pliant crescent]

Link it

[sharp island]

Also do I delete the saved Epoches that are made when it’s training

[pliant crescent]

That literally means auto save is on

[sharp island]

Oh

[sleek nova]

You compare those saved files with your tensorboard graph.

At some point, the tensorboard graph will trend upwards. Identify the step that it does and compare that to the save files, should have one with a step count near it.

Use that file as your starting point for testing your model, itll be at the point of overtraining.

[proper pivot]

Why does it takes so long to convert

80 secs +

[sharp island]

So do I still delete em

[sleek nova]

Dont delete unless you are retraining from a point or you have finished testing and found the one you like best.

[rare goblet]

Ayo? @sleek nova level 6 !!!

[green pond]

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

[viscid moss]

Did u run the first cell?

That takes like 15 minutes

[green pond]

oh no

Ok i will wait for it to finish then get back

[viscid moss]

okkie

[green pond]

i was watching like a 2 year old video so it problably wasn't going to be to correct

[strange imp]

Are there any working kaggle notebooks for training? I was using colabs but I hit my usage limit as soon as I figured out what I was doing

[viscid moss]

https://www.kaggle.com/code/hinabl/multinotebook-ai-public

[glad zealot]

oh hey its mine OwO

[viscid moss]

yep OwO

[brittle wing]

can someone pls help me set up my voice changer? 😭

[rare goblet]

Ayo? @brittle wing level 1 !!!

[viscid moss]

https://rentry.co/VoiceChangerGuide
Check this guide

[brittle wing]

does it work on windows 11?

i cant get it working

ive followed it

[viscid moss]

yep

[brittle wing]

whaaaa

[viscid moss]

let' tal in #🔍│help-w-okada

[brittle wing]

okay

im in vc btw

[viscid moss]

i can't rn

[brittle wing]

i want to load the gui for my pc which is open local from .bat, where's link?

[low lynx]

I'm running RVC Project in Google Colab Pro (Retrieval_based_Voice_Conversion_WebUI_v2.ipynb) for the first time and I'm running into an issue. I see under the cell "Load packaged dataset from Google Drive to /content/dataset" I see this error code "Could not parse variable and value from “/content/drive/MyDrive/dataset/lulu20230327_32k.zip". Expected the line to start with a variable assignment." I am supposed to see "Dataset location Dataset: /content/mydrive/dataset/lulu2023027_32k.zip" with the ability to type in your own .zip file. I ran all the other cells above with no errors and my google drive is mounted. I even saw this error before I began running any cells. I'm using this tutorial video: https://www.youtube.com/watch?v=9wu6LSue_dU

[timid tundra]

Does anyone know how to fix ModuleNotFoundError: No module named 'tkinter'

in python

[clear laurel]

What is the best Collab to use if I want to convert my pre-recorded voice into a set voice from weights.gg

[stoic cloud]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[stoic cloud]

This should help.

[waxen mural]

Im trying to download my converted audio file but theres no 3 dots to choose download

[violet heron]

Did you convert it

[waxen mural]

yeah i can hear the audio file and its definitely converted, but theres no download option

[violet heron]

Using…

[waxen mural]

local version of rvc

[violet heron]

Send a screenshot

[topaz valve]

😭

should I give up trying

[fluid topaz]

imagine using linux💀

[topaz valve]

I hate it so much

do NOT buy a steam deck

[viral flint]

https://pcpartpicker.com/list/sWy2jH is this good for rvc?

[fluid topaz]

bro you forgot the cpu cooler and fan case, i'd not recommend to use the stock cooler that comes with cpu package

[viral flint]

is it bad?

[fluid topaz]

so bad

[sharp island]

@brittle wing hey is this overtraining

[brittle wing]

why did u ping me

[viral flint]

what’s a good cooler

[fluid topaz]

there are 2 kind, AIO watercooling and air cooling, which one do you prefer?

[viral flint]

idrc

[fluid topaz]

well if you looking for budget one, this is good enough

[rare goblet]

Ayo? @fluid topaz level 7 !!!

[sharp island]

Sorry accident

@stoic cloud it’s not connected to runtime but the model is still training what do I do!?

[stoic cloud]

You can reconnect with the runtime by loading the same notebook you were in with the same account, it should automatically connect

[sharp island]

Do I press change runtime type

[stoic cloud]

No, it should just connect automatically when you have the notebook open

If not, try refreshing the page

[sharp island]

I see something saying connecting with three green dots

[stoic cloud]

Yep, that means it is connecting!

[sharp island]

Oh thank lord!

[sharp island]

Also is this overtraining

[stoic cloud]

Graph seems to still be going down

[limber trout]

It's been loading all the time since yesterday

[proper pivot]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

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

[topaz valve]

-linux

pls

[radiant mesa]

how would one remove background vocals like harmonys

[sleek nova]

Karaoke models in UVR

[sharp island]

@stoic cloud do I click export model or index training first?

[stoic cloud]

Index training

[proper pivot]

Hello can anyone help me to zip a folder?

[sharp island]

Export model is red and says export model is not defined

What does that mean

[proper pivot]

https://drive.google.com/drive/folders/1EOY9_E-O9GhbCAOwXicoYRJRS2j4LjYs

Help

Then I will download it

[proud elbow]

train index and model first

[sharp island]

I got an error for export model
Says name “experiment_name” is not defined

[proper pivot]

?

I can't zip it cause I can't download it on my phone

[proud elbow]

tbh you don't need to, just grab the added_*.index and some .pth model snapshots from weights folder

[proper pivot]

Helppp 🥹🥹🥹

[sharp island]

So export model is useless?

[sharp island]

@stoic cloud is there a fix for this?

[sharp island]

And where’s that?

[stoic cloud]

You didn't properly run the notebook in order it seems like

[sharp island]

Huh?

[stoic cloud]

Do you have the model and index file saved?

[sharp island]

I have no clue, do they both end up in the RVC logs zip file

[proper pivot]

@stoic cloud help me

[stoic cloud]

I have not used the colabs before, although you seem to have used them since you do have model maker

[stoic cloud]

when you download a folder, google automatically zips it for you

[proper pivot]

It's my own folder

I can't download it on my phone

Btw can anyone help me to convert?

[sharp island]

@torpid python

[proper pivot]

I'll give you the link to my model

It's been 200 secs and mine haven't finished

[torpid python]

Heyo

[stoic cloud]

Hey hey

[proper pivot]

Help me pls I can't do it since yesterday

[torpid python]

model files (.pth) are in weights folder while the index file are in logs folder

[sharp island]

So is it in the zip folder or not

[proper pivot]

No

[torpid python]

what zip folder

[proper pivot]

Mines not

[torpid python]

You mean a model you downloaded?

[sharp island]

rvcLogs.zip

[torpid python]

Well open it and find out, but my guess would be no

the folder is just "logs"

that sounds like some sort of text logs

[sharp island]

I don’t see one

[torpid python]

are you on local or applio?

[sharp island]

I see an index file

[sharp island]

Local

[torpid python]

what does it say

is it "added..... .index"

[torpid python]

Yes the first file is your index file

[sharp island]

Oh thank lord🙏🏽

[rare goblet]

Ayo? @sharp island level 8 !!!

[sharp island]

What’s the second one I need?

[torpid python]

The .pth file, which should be in a folder called weights

[sharp island]

Yeah I’m not seeing that folder

[torpid python]

do you see an "assets" folder

[sharp island]

No but it seems like the logs folder was in the rvcLogs.zip

[torpid python]

okay go out as far as you can in files

like until you reach a folder with rvc in its name

[sharp island]

Nothing…what abt the npy files

[torpid python]

send screenshots

[torpid python]

No stop looking at rvclogs

there's no reason to do that right now

[sharp island]

They’re the only ones I have

[torpid python]

click this

and send a screenshot of what you see, please

[torpid python]

Okay

[proper pivot]

Can you help me to convert an Ai cover

It tooks me 200secs and I get nothing

[torpid python]

The model, did you download it from anywhere or did you just train it

[proper pivot]

I only use my phone

[sharp island]

Train it

[torpid python]

On rvc disconnected?

[sleek nova]

Do we know why mode collapses occur? I know what they are, but haven't quite found the why of it.

[sharp island]

Yes

[torpid python]

Okay try searching outside of the rvcdisconnected folder for the folders I mentioned just now "weights" or "assets"

[proper pivot]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[sharp island]

Already did none of that i see

[torpid python]

Idk anything about rvc on mobile but my suggestion is you do this on pc instead

[sharp island]

I do see the RVC itself when I look up wights

[torpid python]

What folders do you use

[sharp island]

What

[torpid python]

RVClogs?

Or another RVC thing

[sharp island]

No just I looked it up on google drive

[torpid python]

Yes what do you mean you see the RVC itself

what RVC do you see

[radiant mesa]

which ones i tried a few

or whats the best one

[torpid python]

can you send the link to that google docs btw

[sleek nova]

I use Karaoke 2 from MDX

[sharp island]

https://docs.google.com/document/d/1XuxQYiqEhYrdYeCZRRLrmV_ciMKo0bV-jTCGHu_-5Cc/edit

[proper pivot]

Can anyone help me to convert an ai cover using my model?

[rare goblet]

Ayo? @proper pivot level 3 !!!

[torpid python]

Okay is that zip file unrelated btw

[proper pivot]

I can't do it since yesterday pls

[sharp island]

God dammit

[radiant mesa]

okie thank u

[torpid python]

Like was it there prior

[sharp island]

No? I don’t think so

I was mostly watching over the RVC

[torpid python]

Okay open it and check whats inside

[sharp island]

You mean this?

[torpid python]

No not rvclogs

Mikeysmom.zip

[torpid python]

The one in the image I replied to

Is it unrelated

[sharp island]

That’s where the dataset is

[proud elbow]

the weights folder can be found in the RVC disconnected directory in left pane of the colab page, not in your Drive directory

[sharp island]

Which button

[proud elbow]

yeah, click the folder icon first tho

[torpid python]

Ugh

[proper pivot]

What is this

@torpid python

Input and output path

[proud elbow]

oof, you should have done it while the colab runtime was running

[torpid python]

@proud elbow would it really be there, the guide for rvc disconnected shows the pth and index being in the google drive + why would the index be in the google drive but not the pth + the OG colabs that ran non-local would save in drive too

[sharp island]

The fuck…

[torpid python]

Like his index file is saved in his google drive

why wouldnt his weights be as well

[hoary garden]

is this a help channel for the ai voice changer thing?

[proper pivot]

Help

[sharp island]

Is there a fix or something

[torpid python]

Are you using rvc realtime or w-okada

[hoary garden]

what

[proper pivot]

😫😫😫😵😵

[hoary garden]

im using voice changer client demo

[torpid python]

The voice changer, there's two major ones

what

[hoary garden]

MMVCserver

[proper pivot]

Can voice changer use pth

[torpid python]

Yes it requires pth

[proud elbow]

rerun the colab from beginning, good if you managed to save some preprocessed dataset and training snapshot

[hoary garden]

can you explain

[torpid python]

Ill be honest with you I have no idea what that is

[sharp island]

So I run the dependencies?

[hoary garden]

i thought this server was based on that voice changer

[radiant mesa]

its like

a web server version

[torpid python]

There's two major voice changers (that this server focuses on ig), which are 'rvc realtime voice changer' (this channel) and w-okada voice changer #🔍│help-w-okada

[radiant mesa]

like a hub

[proud elbow]

#🔍│help-w-okada if you mean the realtime voice changer

[radiant mesa]

and you could control it with a another computer

[hoary garden]

alr

[robust mist]

whats the rvc colab

[sharp island]

This shit?

[torpid python]

Okay wait, do you want to like speak into a mic and have an ai voice come out, or put an audio file in and have an ai voice come out

when you say voice changer

[hoary garden]

any of you helpers for wokada?

[torpid python]

just go into #🔍│help-w-okada , scroll up until you find someone with a blue name and then ping them

[hoary garden]

bet

[rare goblet]

Ayo? @hoary garden level 1 !!!

[torpid python]

blue like any shade of blue

[sharp island]

@proud elbow

Ok so I’m running dependencies how long until the pth file show up

Ok uh little update

I clicked the back button found the logs for a brief second until it took me to this

@proud elbow ok I found it

[proud elbow]

yeah that weights folder in the bottom

[sharp island]

What’s the name of the file I need

[proud elbow]

the model checkpoints will appear in the weights folder during model training

[sharp island]

@proud elbow

[proud elbow]

seems rather unethical to keep pinging persons with just screenshots without explaining what it is

or do you mean the error in GPU check part?

[sharp island]

I’m sorry
The reason I did this because I don’t know which is the right one

[proper pivot]

@torpid python

Help

[torpid python]

yes

[proper pivot]

How to convert in hugginfavr

It tooks me 200 secs and nothing happened

The rvc

Help

[sand iris]

what do you need

[slate pawn]

I need help too

[swift lotus]

Is this possible on mobile or nah

[copper pebble]

sup

[copper pebble]

it depends on whatcu want to do

inference is the only thing possible on mobile atm

[copper pebble]

its normal it depends from ur audio lenght and if duplicated the space or not

[slate pawn]

aight so I have this index file that's not working on models I've trained

the moment I use the index file this pops out and the entire model stops working during RVC

[torn aspen]

I get this thing

[proud elbow]

-rvc

[azure marsh]

Suggestion for @torn aspen

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

🎄 Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[rare goblet]

Ayo? @proud elbow level 14 !!!

[proud elbow]

I'd suggest to use one of these RVC forks

[copper pebble]

#🔍│help-w-okada

[rare goblet]

Ayo? @copper pebble level 42 !!!

[torn aspen]

i cant even install or run it 😄

[rare goblet]

Ayo? @torn aspen level 2 !!!

[proud elbow]

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases try this could work well

[torn aspen]

i done this one but it wont launch , i start the "run" it just be visible 1 sec then closing

[slate pawn]

I thought I have a problem with the training it self though, since other index files are working but the ones that I train myself doesn't work

[sleek nova]

Are you using the added or trained index

[torn aspen]

it says that in 1sec i catch it

[copper pebble]

can u send me ur model in dm so i can check it??

btw only added.index is required

[torn aspen]

anything about that ?

is there any video for installation ?

i watch 3 but didn't work

[marble owl]

I have a problem the VRC works good for maybe 3 or 4 times converting after that the pitch changes it gets higher if i convert it more times? Anyone know why or whats happening?

[patent pasture]

hi

I'm working on kaggle, how to upload files input files? Just on colab it was easier

[rare goblet]

Ayo? @patent pasture level 3 !!!

[torn aspen]

hi what is Epochs:

[patent pasture]

oh, I understood probably

[swift lotus]

whats inference?

[copper pebble]

covers

[patent pasture]

How much Batch size in kaggle?

[cold vessel]

was i supposed to extract all files in the start_http

[patent pasture]

16 or 32?

[brittle wing]

This link https://colab.research.google.com/dri...
doesn't work😭

[copper pebble]

it depends on ur dataset

usually 8 is the standard

[cold vessel]

Also could someone give me a quick rundown on how they use it and what i need..?

[patent pasture]

ok, thanks

[cold vessel]

,<

[brittle wing]

what can i do?

[rare goblet]

Ayo? @brittle wing level 1 !!!

[patent pasture]

training started, thanks

[copper pebble]

@modern oxide ya got the colab for voice changing

[modern oxide]

i've been summoned

and if you mean w-okada voice changer then no xd

[copper pebble]

thx anyway

[copper pebble]

https://tenor.com/view/with-this-treasure-i-summon-megumi-mahoraga-gif-17411827983770801229

[modern oxide]

np :3

https://mxdia.giphy.com/media/eR8OEiPdbzQzKrAPHn/giphy.gif

[copper pebble]

running okada its pretty impossible on laptop right?

[cold vessel]

I just need a voice changer..

[rare goblet]

Ayo? @cold vessel level 1 !!!

[modern oxide]

do you have a powerful pc or a laptop?

you can run it locally

[copper pebble]

8gb ram and i7

i wasnt sure thats why i wanted ur confirm

seems pretty low for me

[patent pasture]

Is it normal that 10 epoches takes 100 secs always?

[modern oxide]

no gpu?

[copper pebble]

no :£

[patent pasture]

I thought 2 gpus will be faster

[modern oxide]

zamn, yeah that's not good enough for local :/

[cold vessel]

Nope..>,<

[copper pebble]

dataset lenght??

[cold vessel]

Anything else i can do..

[copper pebble]

thats why i thought too

[modern oxide]

colab, but tbh i never used voice changer colabs, soc an't help with that

[patent pasture]

2:29

[cold vessel]

Can you send it to me in dms

[copper pebble]

minutes??

[modern oxide]

i don't have the link, sorry :c

[patent pasture]

yea

[cold vessel]

damn..

[copper pebble]

well 10 epochs 100 seconds are 1 epoch 10 seconds

[modern oxide]

what gpus do you have?

[copper pebble]

how much epochs did u set??

[patent pasture]

I use kaggle and there 2 are teslas

[patent pasture]

500

[modern oxide]

hmm, weird, it should be faster than colab then

[patent pasture]

just it is strange that it takes only 10 sec on all way

[copper pebble]

watch the tensorboard

[patent pasture]

Are there some
restrictions?

[copper pebble]

wdym

[patent pasture]

when I worked on colab it took 7, 8, 9. But on kaggle it exactly 10 secs

[copper pebble]

weird cuz it should be faster

[patent pasture]

maybe I shell make butch size bigger

[timid tundra]

Does anyone know how to fix ModuleNotFoundError: No module named 'tkinter' in Python

[patent pasture]

do you need a its notebook?

[patent pasture]

pip install?

[modern oxide]

nah, i don't use kaggle

[patent pasture]

it's cool

[patent pasture]

just see https://www.kaggle.com/code/hinabl/multinotebook-ai-public

there are Mainline RVC, Mangio Tweaks, Voice Changer, CoverGen all in one place

[timid tundra]

dont work

[serene geyser]

is there a guide how to make a model itself from audio?

[rare goblet]

Ayo? @serene geyser level 1 !!!

[patent pasture]

do you know how much 1 epoche takes in colab?

[modern oxide]

depends on a length of the dataset. 5 minutes usually take 8-10 secs per epoch, 2-3 minutes take around 4-5 seconds

[patent pasture]

I don't know why does it takes same time?

30 mins and 210 epoches

[patent pasture]

i think it is nerfed or something else

[scenic gale]

Hi, I'm having some trouble and I'd really appreciate some help. I have no coding experience and I'm trying to make a rvc model. When I try to run go-web.bat, it says:

C;\RVC\RVC-beta0717>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
'runtime\python.exe' is not recognized as an internal or external command,
operable program or batch file.

C:\RVC\RVC-beta0717>pause
Press any key to continue...

I'm really not sure what I'm to do, I looked up the error to see if anyone else has had the issue, and I saw someone saying to like, change the command, so I turned the .bat file into a .txt file and pasted it in, but it didn't seem to work. I also tried installing python in hopes it'd fix the issue, but it didn't work.

[inland sphinx]

does this program only work with nvidia gpus?

[scenic gale]

Are you asking me? Because I'm not really sure. My laptop does have nvidia geforce rtx though.

[sand iris]

How'd you download it ?

[inland sphinx]

ah because ive tried it and it doesnt seem to work

[sand iris]

Yesn't

[scenic gale]

I went to the rvc huggingface page and downloaded the version simply listed "beta", i extracted it to my C drive.

[rare goblet]

Ayo? @scenic gale level 1 !!!

[sand iris]

It mainly supports nvidia cards (you can train models & infer)

[inland sphinx]

ive tried many different settings, it doesnt seem to be rendering

[rare goblet]

Ayo? @inland sphinx level 1 !!!

[sand iris]

It also happens to support:

• GPU acceleration on MacOS (bugged from what I've seen), but it supposedly trains & infers
• intel graphics, though training only works on the dedicated cards
• amd, on windows it can infer, on linux it can both infer and train

I'm not too sure on how to get it working on Intel though

[sand iris]

What have you installed right now ?

[sand iris]

Seems to be an older version, but it should be working

[scenic gale]

I see, not really sure why it isn't

[sand iris]

try fetching the latest from here
https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases

[scenic gale]

Alrighty, will do

[inland sphinx]

MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.17b

is there a specific version to use?

[sand iris]

Oh, you want to head over to #🔍│help-w-okada then for that

[astral perch]

I gave up on applio

Im just installing the normal rvc version now

This one has rvmpe as a pitch extractor wich is good

Idk exactly what the differences are tho

[supple ibex]

what sample rate should I use for this model?

this is just a small part of the dataset

[finite galleon]

32k

[patent pasture]

I trained model and now will test

[rare goblet]

Ayo? @patent pasture level 4 !!!

[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

[astral perch]

My RVC doesn't have a weights folder

It's saving the epochs to the D and G file

[patent pasture]

its L

[astral perch]

Nvm found it, it was in the assets folder for some reason

[rare goblet]

Ayo? @astral perch level 8 !!!

[dense moon]

I did everything i could to make my model sound better but it all just end up sounding abstracted

[rare goblet]

Ayo? @dense moon level 2 !!!

[patent pasture]

really, in different rvc uis are different pathes

[dense moon]

It doesn’t help that my character only has 20s worth of voice lines is it..

I even tried duplicating with different effects to maybe make the ai detect it as different pitches but still sounds just the same..

[astral perch]

Is it safe to move my weights folder into the main rvc folder, will it still safe the models

Or does it need to be in the assets folder?

[supple ibex]

thanks

[dense moon]

Is there specific epoch or smth in the training for specifically few voiceline characters or is it really impossible to replicate those types..

[brittle wing]

I get that error too

[proper shale]

check if your folder has spaces or something

[crisp plover]

when you use the index file of a model will that increase the quality ?

[loud raven]

were can i downloadd voice ai

[crisp plover]

https://discord.com/channels/1159260121998827560/1175430844685484042

there

[proper shale]

no, will make result more realistic in terms of accent

[crisp plover]

Thanks 😄

[tulip cloak]

/content/Mangio-RVC-Fork/logs/modelname/3_feature768
Proper Numpy version detected.
Trying doing kmeans 211079 shape to 10k centers.
/usr/local/lib/python3.10/dist-packages/sklearn/cluster/_kmeans.py:870: FutureWarning: The default value of n_init will change from 3 to 'auto' in 1.4. Set the value of n_init explicitly to suppress the warning
warnings.warn(
Init 1/3 with method random
Inertia for init 1/3: 896883.5625
Init 2/3 with method random
Inertia for init 2/3: 890529.625
Init 3/3 with method random
Inertia for init 3/3: 891258.0
[MiniBatchKMeans] Reassigning 128 cluster centers.
Minibatch step 1/82452: mean batch inertia: 30.664289474487305
[MiniBatchKMeans] Reassigning 128 cluster centers.
Minibatch step 2/82452: mean batch inertia: 25.69953155517578, ewa inertia: 25.69953155517578
[MiniBatchKMeans] Reassigning 128 cluster centers.
Minibatch step 3/82452: mean batch inertia: 31.805721282958984, ewa inertia: 25.71434285487554
(10000, 768),256
Training index...
Adding...
Saving files to Drive...
All done! Your index file has completed training.
it gives me this error when doing index
the length of the dataset - 1hr 20 min

anyone know a fix to this issue? the index file only turns out to be only 30 mb when its supposed to be 400mb atleast since the dataset is very long

[proper shale]

yw :)

[proper shale]

well, that happens. iirc you need to change the inner configurations of rvc to fix that

either way, you probably don't need an hour of dataset

[obtuse stag]

does anyone know where that ghost face voice is from?

the one everybody uses

[tulip cloak]

does it affect model quality

[dense moon]

[proper shale]

not necessarily? it just uses a short part of ur dataset for the index tho

[tulip cloak]

okay gotcha thanks

[obtuse stag]

can someone help

[rare goblet]

Ayo? @obtuse stag level 1 !!!

[obtuse stag]

damn does no one actually know about where it comes from?

[proper shale]

idk even what a ghost face is so no

try searching for it in weights.gg or #1175430844685484042

[obtuse stag]

its that famous one that has a raspy voice

[proud elbow]

find and remove this in infer-web.py:

    if big_npy.shape[0] > 2e5:
        infos.append("Trying doing kmeans %s shape to 10k centers." % big_npy.shape[0])
        yield "\n".join(infos)
        try:
            big_npy = (
                MiniBatchKMeans(
                    n_clusters=10000,
                    verbose=True,
                    batch_size=256 * config.n_cpu,
                    compute_labels=False,
                    init="random",
                )
                .fit(big_npy)
                .cluster_centers_
            )
        except:
            info = traceback.format_exc()
            logger.info(info)
            infos.append(info)
            yield "\n".join(infos)

[tulip cloak]

I am using colab so I don't know how that would work

But thank you that looks promising

[proper shale]

dont think u can even alter it tbh

[tulip cloak]

I know hence why I am buying a pc next year

[tulip cloak]

I am reducing the dataset length to 40 minutes

[proper shale]

yeah you don't need an hour

[tulip cloak]

Thank you

[covert sedge]

help, program thinks i didn't run Feature Extraction even tho i did run it

[proud elbow]

tbh you could still make fine quality 1h+ dataset with such that code removed (yeah in local RVC tho)

[tulip cloak]

whats your specs

[proper shale]

the problem is, on colab, you can't

[proper shale]

your dataset wasn't exported correctly, basically

[covert sedge]

oh that's why it got trained faster

[proud elbow]

not my spec, but 3060 12 GB is enough for training

[finite galleon]

You can but need to replace the script after the repo is cloned

[tulip cloak]

okie thanks

[rare goblet]

Ayo? @tulip cloak level 15 !!!

[patent pasture]

Is it good value?

It is 400 epohe

[obtuse stag]

no

[patent pasture]

Just after there are only about 30

[proper shale]

yeah its decent, is this smoothed though?

[patent pasture]

thank you

[proud elbow]

that's quite fine quality, as long there are no mode collapses detectable in 0 smoothness

still you need to test several model checkpoints tho

[scenic gale]

I tried installing the latest version, same error pops up.

Don't really know what to do next.

[sand iris]

huh

really weird, does the runtime directory contain anything?

[scenic gale]

a couple of folders and files

folders:
Lib
Scripts
share
tcl

files:
_tkinter.pyd
tcl86t.dll
tk86t.dll

[sand iris]

What did you use to unzip it ?

[visual pumice]

what would 32k do?

[rare goblet]

Ayo? @visual pumice level 4 !!!

[visual pumice]

what are the differences they make with the models?

i tried v2 at 40k on a dataset on my voice and it sounded horrendously terrible

[viscid moss]

https://rentry.org/rvc_making-models#sample-rate

[brittle wing]

im speaking nth happens

tsunami cat doesnt work for me

[plucky locust]

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

[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

[green pecan]

hey hey, just wanna to ask if there is a tool to easy cut an audio file for training into mutiple parts

[rare goblet]

Ayo? @green pecan level 1 !!!

[proper shale]

you don't need to do that

just use one audio file and u should be good

[green pecan]

okay thank you 🙂

[proper shale]

yw :)

[fossil glen]

for dataset name no capital alphabet no symbols

[sand iris]

Capital letters are fine

They just gotta match

dashes & underscores should be good too

[fossil glen]

its easy gui

[sand iris]

yes ?

[fossil glen]

when i try its getting error if have symbols or capital

[velvet carbon]

where do i drop my pth and index files at in mangio?

[molten pecan]

pth goes in "weights" folder, index goes in "logs" folder

[halcyon flicker]

how come none of the models i tried work?

[molten pecan]

which RVC are you using?

[halcyon flicker]

v2

[molten pecan]

which fork?

[halcyon flicker]

where can i see that?

[molten pecan]

send me a screenshot of the app and I'll let you know

maybe you knew before you download it. the most common forks are Applio, Mangio, EasyGUI

and then there's RVC-GUI which is super outdated and you shouldn't use it

[molten pecan]

oh you are using Colab hm

nowadays Colabs are a little unstable. I'll send you the most stable one

[halcyon flicker]

alright

[rare goblet]

Ayo? @halcyon flicker level 1 !!!

[molten pecan]

Colab link

How to use it

[sudden mango]

türk

birisi

yardım edebilirmi

[gleaming tusk]

rvc doesnt work anymore

[rancid pine]

What's Model fusion under ckpt processing?

Tried to redo training with 32k sample rate

[misty stream]

how would i install mangio rvc on arch linux

[ebon cipher]

how do i get xxx hes not on the thing when i searched it never gave me a link

xxxtentacion

[copper pebble]

search on weights.gg

[ebon cipher]

i got it

i just copied the link on download

[copper pebble]

well uve done then

ya wanna do covers?

https://discord.com/channels/1159260121998827560/1167844377046040769 follow this for covers

[ebon cipher]

it says unkown

[ebon cipher]

im trying to get x as the adlibs for my song

[static gyro]

i cant get appio to work

[violet heron]

That is not RVC, that is w-Okada

Did you select your files for the model?

[gleaming tusk]

yep

[violet heron]

Also I recommend turning index to none

[gleaming tusk]

you can rvc inscription

[violet heron]

Do the other models work?

[gleaming tusk]

yes

but so rapy

what is w-okada

[violet heron]

The voice changer your using

“Voice Changer Client Demo” is also called w-Okada

[gleaming tusk]

what about your tool gimme a look

[violet heron]

Can’t rn not at my computer

Are you sure you uploaded the .pth and .index file on the edit menu for your model?

[gleaming tusk]

its stock

factory

[violet heron]

So what do you need help with?

Adding a model?

[gleaming tusk]

Voice without distortions

Gimme a look your tool

I will try

[violet heron]

It’s going to be different since I have different hardware then you

[gleaming tusk]

Gimme a link

And

whats a difference?

[violet heron]

Some settings work better or worse on certain hardware

[gleaming tusk]

It will be fine on my own

Gimme a try

[violet heron]

If you need any help you can also refer to these links #🔍│help-w-okada message

[gleaming tusk]

whats the better tool?

[violet heron]

Your using it

[gleaming tusk]

why you are using different?

[violet heron]

I don’t use different

I use the same program as you

It’s just that different settings work better on different computers

[gleaming tusk]

I want make my voice more natural

[violet heron]

Tune or Extra

[gleaming tusk]

On max value?

[violet heron]

No

Just try different ones and see what works well

[gleaming tusk]

do you have trump model?

[rare goblet]

Ayo? @gleaming tusk level 3 !!!

[gleaming tusk]

thx babe so much

[violet heron]

Check the website https://weights.gg

[gleaming tusk]

what about chunk?

[violet heron]

You can also adjust that to see what sounds better/worse

[rare goblet]

Ayo? @violet heron level 80 !!!

[gleaming tusk]

thereshold?

threshold

[violet heron]

I wouldn’t really mess with it

[gleaming tusk]

thx you babe

shall we play together

[polar siren]

trying to train a model for the first time on rvc (I always used kits.ai before) and I just wanna know if this is supposed to happen? and how long this goes on for?

for mote info, Im training a model for 50 epoches (since the audio im using is only 1:30) and have it so it's saving every 25 epoches

[astral perch]

i keep getting this error when trying to download tensorboard ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow

[violet heron]

A lot

Depends on your GPU tho

[polar siren]

ok, so it is normal and it's not broken?

[hollow torrent]

What technically does one have to do to try to replace Hifigan with another gan? I'm trying to see if replacing hifigan with Bigvgan(https://github.com/NVIDIA/BigVGAN) would yield an improvement. But I'm new to this AI Python stuff, so I'll probably rely on GPT to help me or give up in the end.

[misty stream]

so how do i install mangio rvc on arch linux

[sand iris]

get python 3.10 from the aur

setup a python virtualenv with python3.10 -m venv /wherever/the/hell/you/want & activate it via source /wherever/the/hell/you/want/bin/activate

clone the code from github via git

download the proper torch release (refer to https://pytorch.org/)

then install the requirements via pip install -r requirements.txt (assuming you cd'ed into the repo earlier)

make sure you have aria2 installed, and from the code's root, run ./tools/dlmodels.sh

then all that's left is running python-infer-web.py

I'd suggest using the original RVC repository though, mangio hasn't been touched in a long while

[calm crown]

Does anyone has an applio colab link that works?

[misty stream]

do i specifically need torch 2.0.0 as stated in the requirements.txt

[rare goblet]

Ayo? @misty stream level 1 !!!

[sand iris]

nah

tested & works fine on 2.1

[violet heron]

No

[molten pecan]

<#1175615544037212200 message> not Applio but still pretty good

[half cove]

I mean I have one

But gatekeeping is real

[violet heron]

that works

[half cove]

Yea

[old oracle]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

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

[brittle wing]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[rare goblet]

Ayo? @brittle wing level 2 !!!

[azure marsh]

Command has expired

[brittle wing]

-help

[azure marsh]

Command has expired

[calm crown]

Thank youu

[rare goblet]

Ayo? @calm crown level 1 !!!

[brittle wing]

-train

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[blissful kraken]

how to use hina kagle mod, i cant seems to run any code

ohhh its okay, i got it, i have to click edit

[brittle wing]

-help

[azure marsh]

Command has expired

[real rock]

So I havent used rvc for a while, and I cant seem to get it to come out of my voicemod, but sometimes I can hear myself talk but voicemod doesnt pick it up, any suggestions for microphone settings for rvc?

[rare goblet]

Ayo? @real rock level 1 !!!

[violet heron]

Don’t use Voicemod

[real rock]

dang well then idk how id get it to work :/

is there another way to get it to work?

got it working, ive never had issues with voicemod before, but i guess for now, i wont use it

[misty stream]

is there a version of this script that works on linux

[wooden plinth]

you guys the gasoline that i use on my computer to run rvc isnt strong enough, which gasoline should i use

[idle lotus]

voice models end up really choppy on discord but on like cs2 and steam games they're fine? I'm using a VB virtual audio cable. Someone please help me? Also im using the nyanners by dan model

[misty stream]

have you tried diesel or camping gas

[wooden plinth]

oh daim thats smart, you sure know a lot about computers

[dense moon]

My voice model just sound like pure static even tho theres not really much background noise in the audio of the character i used, but the character only has 20s worth of words that they can only speak, is it that which is affecting the ai to not be able to process it well?

The song i wanted my voice model to sing was in japanese and that didn’t sound anything coherent lol, i thought its because of the language but then when i switched it to sing an english song, it still is just pure static..

[misty stream]

-help

[azure marsh]

Command has expired

[tepid cave]

use prem gas

[wooden plinth]

bruh you think gas grows on trees 💀

[rare goblet]

Ayo? @wooden plinth level 1 !!!

[tepid cave]

it'll fix it

[old oracle]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[static gyro]

is there a spectrogram thats free?

[viscid moss]

Spek or Ilaria Audio Analyzer

[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

[blissful kraken]

i am using keggle and it always turn off after 30 minute, what should i do so its still keep running

[rare goblet]

Ayo? @blissful kraken level 3 !!!

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

[blissful kraken]

i am using keggle and it always turn off after 30 minute, what should i do so its still keep running

[proven hill]

BRAND NEW FEMALE VOICE MODEL!!

Female e-girl sexy voice model for Okada (RVC V2) is now avaliable on my Ko-Fi!

Get it now here! https://ko-fi.com/s/734d5f276c

Or find other female models (including free ones) here: https://ko-fi.com/ilariaowo

20% discount if you DM me!

[spring fossil]

hi, when i convert the audio to the ai voice it says to download it using the 3 dots but there arent 3 dots

[blissful kraken]

its robotic

[proven hill]

its not robotic at all?? 💀

[rancid dove]

Anyone know why my rvc covers aint working

it makes a error

File "C:\Users\indyk\OneDrive\Desktop\AI COVERS\RVC-GUI-pkg\rvcgui.py", line 115, in vc_single
audio = load_audio(input_audio, 16000)
File "C:\Users\indyk\OneDrive\Desktop\AI COVERS\RVC-GUI-pkg\my_utils.py", line 19, in load_audio
raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail)

here is what its saying

@proven hill

do yk why

[proven hill]

@true vortex

[true vortex]

audio file not supported

[proven hill]

thanks i was in the bathroom

oh yea ffmpeg

[sand iris]

or ffmpeg not working somehow

[true vortex]

codec of the audio file is not supported

[proven hill]

you can use this tool to convert the file: https://huggingface.co/spaces/TheStinger/Ilaria_Converter

[eternal peak]

welp my applio broke since I havent used it for months

whats the best local RVC rn?

[brittle wing]

Can someone help, the audio file when i drag it doesnt work like the file is unsupported 😭/

[blissful kraken]

i am using keggle and it always turn off after 30 minute, what should i do so its still keep running

[hasty frost]

Does someone know how to make a good acapella?

[crude anchor]

Could it work with CPU to train model

[sand iris]

Technically yes

Now will that be slow ? Yes, really slow

[crude anchor]

When i training the model , when i click the last option to train the model it is not working .the output is same like in your video but there is not have any file saved in weights folder .

[finite galleon]

But doesnt this use ffmpeg as well

[sand iris]

uses it online on hugging face machines

[finite galleon]

Ya

But still ffmpeg no?

[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]

How is training going?

Total epochs 150. Dataset length 38 minutes. Current Training epoch 49

[proper shale]

Seems to be going alright

[hushed osprey]

Greetings
I am having difficulties running an AI voice model
The native voice client interface won't load
I am on a windows PC with an NVIDIA GPU, using the MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.17b.zip version

The AI is set up, but the interface won't load, and the Google browser interface I'm using isn't working either

The Google interface loads, but it doesn't have any sound

[proper shale]

Does CMD say anything?

[hushed osprey]

Well yes

A lot actually

[wide parcel]

what does model ID do

[hushed osprey]

But Idk how to share it without also sharing the IP and name

@proper shale & @wide parcel

[wide parcel]

voice changer?

#🔍│help-w-okada ?

[hushed osprey]

Yea

[hushed osprey]

Alright

[proper shale]

No errors here

[proud elbow]

keep going, also watch the g/mel and g/kl as well

[rancid pine]

How's this?

[proper shale]

Looks decent

[rancid pine]

I figured out the ckpt processing, but with different tensors I can't do much

[latent kettle]

how is training going? 115 epochs

[proper shale]

going well

[latent kettle]

Need more epochs? Or i can stop it ?

[proper shale]

keep going

[latent kettle]

When I have to stop?

[proper shale]

wait if ur training on RVC Disconnected just make sure u havent trained for more than 2 hours lol

and export ur model

Anyways, I don't think you need to stop anytime soon

[latent kettle]

I'm using RVCdisconnected. And training is on going form 2 hours and 50 minute

[proper shale]

i recommend stopping rn and exporting model

[latent kettle]

Can you please explain when I should stop it ?

I want a best models it don't matter how long the training is going

[proper shale]

literally right now

and continue later

[latent kettle]

How to analyse tensor board?

[proper shale]

-overtrain

[azure marsh]

Hey, @latent kettle!

👇 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

[latent kettle]

Sure

[latent kettle]

But I can't understand this! Can you explain it in simple terms?

[proud elbow]

yeah just keep going, though 2 mode collapses spotted, btw if you're training locally and have some spare VRAM you can test infer several model checkpoints right now

[proper shale]

rvc disconnected

so no

[proud elbow]

the index file and model checkpoints can be downloaded while the training is ongoing, so yes

[half cove]

Yea you gotta open the colab files and download through there

[proper shale]

_ _

oh

yeah especially considering Huggingface exists

[brittle wing]

Guys im using mp3 and i have the audio but it wont let me upload it

What do i do

[earnest jasper]

the colab im using disconnect automatically

can anyone send the ai cover gen colab link

pls

[proper shale]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[earnest jasper]

thank you soooo much 😻

[rare goblet]

Ayo? @earnest jasper level 1 !!!

[stray iris]

Hi there, can someone tell me whats defference between Applio and RVC V2? I already understood that Applio just fork of RVC, but what general purpose of each? Which are better for what?

[proper shale]

Both use RVC

RVC stands for Retrieval-based Voice Conversion, allows you to train voice models to convert audio files of a voice to another voice

[stray iris]

I mean, I opened a guide -> https://docs.aihub.wtf where see two options: one cloning guide for Applio and one for RVC V2. So I just dont understand, which I should use, whats difference between this programms? Maybe some is more userfriendly or some need more powerfull hardware?

[rare goblet]

Ayo? @stray iris level 2 !!!

[proper shale]

"RVC V2" isn't it's real name to be honest. Dunno why they named it that in the guide. It's the original RVC. v2 is just the new pretrains and stuff.

Applio's a fork, meaning it has changes to the original program. Applio has an easier to understand GUI, same requirements as original RVC.

[stray iris]

oh, thanks a lot

[proud elbow]

"RVC V2" is the model architecture, that also allows 32k target sample rate and its corresponding pretrain base model

[proper shale]

you're welcome ;)

[agile storm]

-kagge

-kaggle

is there a training notebook for kaggle?

[proper shale]

you don't need Kaggle anymore tbh

[agile storm]

cuz I dont really understand how do I use multinotebook one

[agile storm]

why

[proper shale]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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

[proper shale]

These

[agile storm]

colab is worse, isnt it?

[proper shale]

...no?

they're a bit more user friendly ngl

[agile storm]

its like giving you worse gpus

and less time

[proper shale]

free t4 and 4 hours

resets every day

don't think it's bad

[agile storm]

when kaggle is 30 hours 2x t4

[rare goblet]

Ayo? @agile storm level 1 !!!

[proper shale]

30 weekly hours

[agile storm]

ik

[proper shale]

7 times 4 is 28

just saying

its not that big of a difference, and colab is just easier

[agile storm]

but when I train a model I want it to be training more than 4 hours straight, not stopping and waiting until I can train it again

[proper shale]

¯_(ツ)_/¯

[agile storm]

and I am not sure if I can continue training on this colabs, can I?

[proper shale]

you can

[agile storm]

on every single of them? and which one is better to use then

cuz I've used colab only three times

and every time it was svc

now I am trying to get used to rvc

[proper shale]

RVC Disconnected and EasyGUI specifically are for training, RVC Disconnected being training only

EasyGUI can infer too

[agile storm]

okay, thanks. I guess ill use easygui then

[agile storm]

oh and also, can colab work somehow when I turn my pc off?

there is this feature on kaggle

[tribal thicket]

how do i upload a model on huggingface

[proper shale]

well no, u have to keep the tab open at least

[cerulean aspen]

what do i do when it says

"[W CUDAGuardImpl.h:124] Warning: CUDA warning: misaligned address (function destroyEvent)"

[sand venture]

Hello ! I try to run any version (RVC, DDSP,So-Vits) in realtime but no go... Can I get help on mac?

[proper shale]

Do you have a Mac with the M series of chips

[sand venture]

I have some macs, but also the macbook air M2

i would like it to run on my i7-13700KF thunderbolt audio

which runs Ventura as well

the i7 has a AMD card, as no more nvidia is used in Macs... RX6600XT 8gb

[rare goblet]

Ayo? @sand venture level 1 !!!

[sand venture]

sorry? do i need to do something i forgot?

[sand venture]

is there some i forgot about the bot message?

[proper shale]

yeah use that

[proper shale]

nah its just a lvl up message

[sand venture]

is it good enought to do realtime with a pth model?

[proper shale]

yeah... it is

[agile storm]

well I am currently training my model using colab, but still, how do I train on the multinotebook kaggle? there is a link but its no longer working

[crude anchor]

Is it rvc need gpu to train voice?

[rare goblet]

Ayo? @crude anchor level 1 !!!

[proper shale]

yeah

[proud elbow]

yes, it needs dedicated GPU to train, otherwise it could run much slower like trying to run modern games using CPU/software rendering only

[proper shale]

I don't think there's an active Kaggle notebook iirc

[agile storm]

thats sad

theres no headless on colab

[crude anchor]

I tried with cpu but at last the file is not saving . The terminal also not showing anything ,just like stucked

[tribal thicket]

how do i upload a model on huggingface

[rare goblet]

Ayo? @tribal thicket level 2 !!!

[tribal thicket]

someone answer

[proper shale]

-huggingface

[azure marsh]

Suggestions for @tribal thicket

huggingface.co upload Guides

English guide

Link: https://rentry.org/fdg_guide_newer
Credits: FDG

Spanish Guide

Link: https://rentry.org/AI_Hub_guide_esp
Credits: Julia (ailen2091)

Hindi guide

Link: https://rentry.org/AI_hub_guide_hi
Credits: Enes

Turkish guide

Link: https://rentry.org/AI_hub_guide_tr
Credits: Enes

[sand venture]

fine to me. It supports my UAD soundcard too. Is there a step by step on how to get the real time? I can get the web gui running, but not the other menu with the input/output for real time. And thank you for those answers already.

[proper shale]

oooh i thought you were using W-Okada

[tribal thicket]

what about the index?

[proper shale]

...just zip it with the .pth and uplosd

[tribal thicket]

ok

[proper shale]

I'd recommend using that tbh, idk about realtime on Mac

if it works well

[sand venture]

i have tried the pre built binaries, only one "works", but i get a lantecy sooo impossible, and all is choped/distorded

[proper shale]

With the M2 it should work fine tho

[sand venture]

i would be sooo happy, i think binaries support other models

but no, i tried with headset to be sure of larsen effect

[sand iris]

what's the deal here ?

Oh the rvc realtime thing ? Yeah that thing is beyond buggy

you might want to use voice-changer for real time

[sand venture]

i have the sovits realtime gui running, then i can select a model, a folder for config file, then infer , and app exits... some about a config.yaml

[sand iris]

Don't bother

I can't even import models here, before it straight up didn't launch

It only supports windows

[sand venture]

what would be your advise?

[latent kettle]

After how much time colab resources (GPU) are refilled?

[rare goblet]

Ayo? @latent kettle level 13 !!!

[sand iris]

https://github.com/w-okada/voice-changer/blob/master/README_en.md

There's a guide which can be found at https://rentry.co/VoiceChangerGuide , though I'm unsure if the instructions are right for apple devices

It'll work, but I don't think it's gonna work as good as it could with minor tweaks

[sand venture]

I think I have tried that github on M2 and had errors... will try now

[rare goblet]

Ayo? @sand venture level 2 !!!

[blazing salmon]

i just tried okada from github and after i did a conda env : (okada) macbook@MacBook-Air-de-Macbook voice-changer-master % sh start_docker.sh
VC Client start...(with gpu)
hostname: illegal option -- I
usage: hostname [-f] [-s | -d] [name-of-host]
start_docker.sh: line 22: docker: command not found

same with the other start2.sh