✨│ai-help

[proper shale]

¯_(ツ)_/¯

you do you ig

[frozen sandal]

CPU makes a 3 minute song take 12 minutes to inference

[molten pecan]

Then I'll update the guide

[frozen sandal]

it is that bad

[proper shale]

If you're making covers I'd go with AICoverGen with WebUI

-colab

[azure marsh]

All working colabs / spaces ☁️

• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• 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

Credits to Faze Masta and Antasma for compiling these links.

[molten pecan]

It doesn't have mangio-crepe

[frozen sandal]

⬆️

[molten pecan]

[frozen sandal]

a lot of those are out of date

[proper shale]

Who uses Mangio-crepe?

Lol 💀 rvmpe is just better

[molten pecan]

I do 😭

Quality wise is not. Speed wise it is

[proper shale]

rvmpe is better both in quality and speed

the own research paper said it

[proper shale]

aicg is up to date

literally tested it like 3 days ago, its working lol

[frozen sandal]

aesthetic RVC is the best one and isn't even on that list imo

[molten pecan]

Maybe because mangio-crepe is more prone to artifacting. I think that's what the paper was referring to

[proper shale]

pitch accuracy in general was better iirc

[molten pecan]

Ask @true vortex

[frozen sandal]

i'll try with the aesthetic one, to check if it's an issue with ilaria or with RVC in huggingface in general

[proper shale]

😭

[proper shale]

Probably in general since I think most people don't use GPUs and then devs don't account for it which is kinda meh

[frozen sandal]

i need a GPU based inference, cuz otherwise it takes too long and sometimes i'll wait 10 minutes for an inference just to get an error in the end

[rare goblet]

Ayo? @frozen sandal level 3 !!!

[frozen sandal]

ain't nobody got time to waste nowadays

update, with aesthetic it worked

the problem is with ilaria only, the person who made it should check it

[finite galleon]

Dawg rmvpe is faster

[molten pecan]

I know

[uncut pelican]

um....

what does it mean

i can't handle it anymore

i hate it

i hate it

[placid talon]

is there a guide to the webui by Hina, cant find it

[fossil glen]

for folder name no capital alphabet

[violet heron]

Did you put the dataset in before or after

[summer glacier]

when im making my voice model it gets stuck on INFO:torch.nn.parallel.distributed:Reducer buckets have been rebuilt in this iteration.

[rare goblet]

Ayo? @summer glacier level 3 !!!

[fossil glen]

screnshoot

[summer glacier]

hold on lemme reopen it

[sand iris]

Takes a while

Depends on hardware and dataset length

[summer glacier]

its been like that for 15 minutes

my pc is pretty good

[sand iris]

pretty good as in ?

also how much time worth of audio files did you give it ?

[summer glacier]

its decent

7 minutes

[sand iris]

what's your GPU ?

[summer glacier]

3050

rtx

[sand iris]

the 8GB or 4GB variant?

[summer glacier]

uh

how do i check?

[fossil glen]

taskmanager

[summer glacier]

this is what i see

dang my usage

im training it rn seeing if it will get the error

[sand iris]

huh it's unexpected then

Remember what you put for the "batch size" setting ?

It shouldn't take over like 2 minutes at the very worst to display something else

[summer glacier]

oh

things stuck on this

[fossil glen]

still stuck?

[brittle wing]

hey guys, can you use rvc with a gpu like NVIDIA GeForce MX550? looking at that new laptop

[fossil glen]

yes its have cuda

[fossil glen]

but i dont know about support driver

[brittle wing]

nice, but what does all of it mean?

and what's a support driver

[fossil glen]

you will know if you do it

i never use local pc

[brittle wing]

me neither

wait i use local pc lmao

[rare goblet]

Ayo? @brittle wing level 2 !!!

[robust quartz]

can i ask if i can remove the models i download? and is it on the folder of mmvc on the models i uploaded?

[fossil glen]

ss?

[brittle wing]

why my mic repet what my friend say

[dry flume]

why do all the voices (that even specify for singing) seem extremely monotone, is there something ive completely scuffed up?

[compact sentinel]

Guys, wanted to know exactly how I would go about doing ai vocals

I downloaded a model of ed sheeran's, but im not sure where to go from there. What tools should I be using?

[proud elbow]

I think RVC uses PyTorch that doesn't support CUDA versions older than 11.x

[proven hill]

is this ilaria rvc?

[frozen sandal]

yes, that happened when trying to update the duplicated thing to a new one with GPU support

on huggingface

[proven hill]

did you pay for it??

[frozen sandal]

yes I did

[proven hill]

oh wow, kinda unexpected this would happen

1s

what happened is very simple

you got the exact same error as this person:

https://huggingface.co/spaces/TheStinger/Ilaria_RVC/discussions/2

and i never thought someone would enable a GPU

the issue is, everything is coded and optimized for a CPU usage, using a GPU would require the training stuff wich are missing to preserve space and time

[frozen sandal]

I had to enable a GPU because without it, it would take like 10 minutes to inference a simple 2 or 3 minute audio, even duplicating the space, and risking waiting 10 minutes to get an error in the end was just too frustrating, so GPU at least would make the process faster

[proven hill]

i was already working on a local version with GPU support

[proven hill]

i cant even test this specific issue since they dont accept paypal
im working on the GPU support

[karmic pebble]

So i cant use a GPU due to colab usage limits, and the colab apparently needs a GPU to continue training, tf?

[frozen sandal]

i'll be looking forward to it, in the meanwhile i'll be using the aesthetic rvc space

[brittle wing]

Hello guys I'm getting error while running easy gui.bat

C:\Users\mayan\OneDrive\Desktop\RVC\Mangio-RVC-v23.7.0>C:\Users\mayan\OneDrive\Desktop\RVC\Mangio-RVC-v23.7.0\runtime\python.exe EasierGUI.py --pycmd C:\Users\mayan\OneDrive\Desktop\RVC\Mangio-RVC-v23.7.0\runtime\python.exe --port 7897
Traceback (most recent call last):
File "C:\Users\mayan\OneDrive\Desktop\RVC\Mangio-RVC-v23.7.0\EasierGUI.py", line 28, in <module>
from my_utils import load_audio, CSVutil
ImportError: cannot import name 'CSVutil' from 'my_utils' (C:\Users\mayan\OneDrive\Desktop\RVC\Mangio-RVC-v23.7.0\my_utils.py)

[proud elbow]

EasierGUI is broken and there's no new update on Mangio

[brittle wing]

Okay sir then any other way to get easier webui on local system like YouTuber showing easy webui on Google colab ?

Very confusing here xD

[crimson wraith]

my PC only has 8GB of RAM.

Will inferencing and merging run faster if I upgrade to 32GB of RAM vs just 16GB of RAM? Or is it not really throttled by RAM? i don't have a GPU so i dont plan on training.

[haughty quiver]

how do I use voice models that have various voices in them?

[copper pebble]

u dont even need to do it locally

for inference

there are a lot of good option online

https://discord.com/channels/1159260121998827560/1175615544037212200

like this is on colab

https://discord.com/channels/1159260121998827560/1167844377046040769

theres also this that is on huggingface

[brittle wing]

What UVR models are used on Ilaria UVR?

[crimson wraith]

I'm using the one on colab right now but have a lot of locally saved models and want to do merging. I figured it would be easier/faster to do it on my own computer? Maybe that's wrong, i dunno.

[rare goblet]

Ayo? @crimson wraith level 2 !!!

[brittle wing]

Wow i felt like I was so close

I did everything on RVC Disconnected but the huggingface link doesnt work

if anyone has some time to help that would be great

[karmic canopy]

is it normal that it stayed for that long?

[onyx canyon]

my voice changer keeps being choppy and buffers hard

[peak tendon]

Hey any solutions for Mac m1 ?

[rare goblet]

Ayo? @peak tendon level 2 !!!

[peak tendon]

With my Mac I have excactly this issue

[lament eagle]

overtraining?

[peak tendon]

How to fix this

[karmic canopy]

i have no idea

[glad zealot]

no keep going

[fallow lava]

gradio.exceptions.Error: 'index -1 is out of bounds for axis 0 with size 0'

Does any of you know what this means?

The collab version stopped working for me as well.

Here's proof that it used to work when I linked a YouTube video to this version.

For additional context, I'm using the Hina Mod

[peak tendon]

Maybe conda is missing ?

[karmic canopy]

idk how to do that

[peak tendon]

I’m trying install anaconda

[karmic canopy]

oka

[peak tendon]

its loading and loadin and loading but why on my mac???

[peak tendon]

Any help please

[west hazel]

omg

are u running on mac

use the mainline version

[peak tendon]

yes

[peak tendon]

mainline version?

what do you mean

[west hazel]

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

use this

[peak tendon]

oh and its working for mac also?

[west hazel]

ik it says Nvidia but download that one

and then run sh run.sh from directoru

it's a new build of rvc so weights are stored at assets/weights

the mangio fork is only for training lol from my testing

idk why the main version doesn't train

[peak tendon]

oh okay so i will try. it...

[west hazel]

allg

[sand iris]

pstt

[west hazel]

what kinda Mac do u have

[peak tendon]

danke thank you i will try it now on mac

[sand iris]

you might want to download the code from the github instead

[peak tendon]

M1 Max

[peak tendon]

?

[west hazel]

that's fire

[sand iris]

the nvidia release contains a lot of windows files that are just not needed

[west hazel]

yet its the only download

with Mac support also

[west hazel]

💀💀

[sand iris]

"download zip" instead

[peak tendon]

where can I find the download of this??

[west hazel]

ye listen to that guy

[sand iris]

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

[west hazel]

he's a mod

[peak tendon]

i cant see any download https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

and where to store the weights here?

[west hazel]

assets/weights

[peak tendon]

and my audio file which i want to convert?

[west hazel]

copy as path

anyways I'm done helping bye

[peak tendon]

thank you i will try and report

[sand iris]

Good call lol I was looking for it

https://cdn.discordapp.com/emojis/790214199544512573.webp?size=48&name=kek&quality=lossless

[west hazel]

windows moment

[peak tendon]

last question what to choose here for mac?

[rare goblet]

Ayo? @peak tendon level 3 !!!

[west hazel]

read the last clause

[peak tendon]

ok thanks

[peak tendon]

what is it doing? i have already my pht to use for my audio

[sand iris]

downloading needed files

[peak tendon]

ValueError: mutable default <class 'fairseq.dataclass.configs.CommonConfig'> for fie(b(b(ba(ba(ba(ba(b(ba(ba(b(b(b(b(ba(b(b(b(b(b((ba(((((((((base) @MBPvonEmanuele Retrieval-based-Voice-Conversion-WebUI-main %
I am getting this error now

after executing the run file

[sand iris]

Oh boy

Yeah

[peak tendon]

cant execute it?

[sand iris]

Basically any newer than python 3.10 just breaks

you seem to have a newer version

[peak tendon]

and how to downgrade

[sand iris]

Unsure on how to do that on mac without causing trouble

Have you ever heard of brew ?

[peak tendon]

yes

[sand iris]

Okay great. can you brew install python@3.10 ? Unless either of 3.8, 3.9 or 3.10 are already installed

[peak tendon]

I did thanks, how to check which version is installed?

[west hazel]

there's no runtime in code

[sand iris]

ya I'm aware

[west hazel]

lol

[sand iris]

It sets up a python virtualenv with the default system interpreter

[west hazel]

trolled moment

[sand iris]

or if it can't find it it tries to install 3.8

but anything than 3.10 just breaks one of the dependencies

[west hazel]

fairseq

hence just download the other one but each their own

[peak tendon]

oh no I get always 3.11 python -V
Python 3.11.5

how do uninstall

[sand iris]

brew list | grep ^python

[sand iris]

you don't need to uninstall it

better idea to tell rvc to use a secondary one

[west hazel]

do you code anything outside of rvc with python @peak tendon

[peak tendon]

learned in school years ago so no

[west hazel]

install 3.8 then

[peak tendon]

brew list | grep ^python
python@3.11
python@3.8
python@3.9

[sand iris]

3.10 works and will be supported for signifcantly longer

[west hazel]

remove 3.10 if u wanna cuz LOL

[sand iris]

Oh 3.9 time then

[west hazel]

clearly not

[peak tendon]

but do i need to uninstall my current one?

[sand iris]

they're using 3.11 rn

[west hazel]

it's already in your system

just redirect rvc to it

[sand iris]

I've installed this times and times again on 3.10

[peak tendon]

ah, and how?

[sand iris]

fairseq breaks on 3.11+

[west hazel]

that...
I have never tried since I have the runtime

[peak tendon]

im sorry for my dumb questions, im learning more and more with you

[west hazel]

it's ok bud

[sand iris]

which one ?

[peak tendon]

but now i need to select 3.8 but how

[west hazel]

at least you're not one of those dumbasses that think you're entitled to answers

[sand iris]

you'd have to edit the run.sh script

[west hazel]

ok

risky move here

how about I send you the runtime folder

and you just paste it in the install

see if that works

[peak tendon]

$ brew uninstall -f python python@3.11 this is not working

[west hazel]

I'm brushing my teeth lol hol on

[sand iris]

... or delete the .venv folder, edit the run.sh script and replace the following

-  python3 -m venv .venv
+  python3.9 -m venv .venv

[peak tendon]

after executing the first line with "-" and running again the run.sh i get following error
ValueError: mutable default <class 'fairseq.dataclass.configs.CommonConfig'> for field common is not allowed: use default_factory

[sand iris]

oh no that's not how it's supposed to go x)

[peak tendon]

😦

[sand iris]

first time you launched run.sh it created a .venv directory in the RVC install dir

First, that one must be deleted because it relies on python 3.11

[west hazel]

[peak tendon]

so i delete the venv.sh file from directory?

[sand iris]

Are hidden files/folders displayed ?

[peak tendon]

ah yes now i can see the folder

so i delete the folder right?

[sand iris]

yup !

[peak tendon]

and now?

run the run.sh?

[sand iris]

edit the run.sh file first

[peak tendon]

what shoudl i dedit

[sand iris]

find the line that contains :

python3 -m venv .venv

[peak tendon]

yes

[rare goblet]

Ayo? @peak tendon level 4 !!!

[sand iris]

yeah !

[peak tendon]

so i delete this line?

[sand iris]

now replace that same line with

python3.9 -m venv .venv

[peak tendon]

done

[sand iris]

great

save, and now if you run the script again, things should just work

[peak tendon]

okay now its downloading lets see. now it should work everything?

[sand iris]

yeah

[peak tendon]

thanks so much, i will report 😄

you are the developer of that?

with that line now its using phyton 3.9? only to understand 😄

[sand iris]

Nope! (replying to are you the developer of that)

[sand iris]

Yeah, since you have that installed

[peak tendon]

ah great learned agian something new 😄

[sand iris]

By default the script does this :

If python is installed (any version), use it
else, download 3.8 and use that

[peak tendon]

ahhh so it was my error to install manually the newest phyton

after the skript the web browser should open right? and than everything should work

next issue:
eason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

[sand iris]

I think it's supposed to open yeah

[peak tendon]

what should i do now..

[sand iris]

brew update && brew upgrade

[west hazel]

lol

[peak tendon]

Already up-to-date.

it says

i could cry

ImportError: dlopen(/Users/e/Documents/Dokumente - Emanuele’s MacBook Pro/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <590C66FA-DB62-3794-848C-4DC3F9EA2E70> /Users//Documents/Dokumente - Emanuele’s MacBook Pro/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

this is the error message

[west hazel]

try putting it on desktop

[sand iris]

Oh yeah spaces in folder names is never good

[peak tendon]

ModuleNotFoundError: No module named 'soundfile'

[west hazel]

well

youre getting somewhere

its a different error now

[peak tendon]

potato recommended to use this nvidia version here https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases/tag/updated1006v2
I will try it again i hope no issues comes

[sand iris]

if you compare the run scripts they are gonna be the same

[west hazel]

except the other one has a built in runtime 🥴🥴🥴🥴

[sand iris]

...for windows

windows x86_64 in fact

[west hazel]

oh well

[sand iris]

With the pytorch package for nvidia

[west hazel]

it works on my mac

[sand iris]

so nothing that's ever going to be useful

It's the same thing as running from the git repo, with added 5GB+ of bloat

[west hazel]

mmhmm

[sand iris]

run script is an exact match

[peak tendon]

but why do i get this error:
File "/Users/xy/Desktop/Retrieval-based-Voice-Conversion-WebUI-main/infer/modules/vc/modules.py", line 7, in <module>
import soundfile as sf
ModuleNotFoundError: No module named 'soundfile'

if i try with the other version you mentioned

should i run sh run.sh or sh ./run.sh?

[sand iris]

both do the same

source/bin/activate && pip install -U soundfile

[peak tendon]

now i am executing again the run.sh skript with sudo (otherwise it would not work to execute the nvidia one)

[sand iris]

no!

[peak tendon]

zsh: no such file or directory: source/bin/activate

[latent kettle]

I'm installing AiCoverGen on my local system and getting this error any solution to fix this ?
Failed to build fairseq
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects

[sand iris]

are you doing this from the RVC folder ?

[peak tendon]

ah no should I

[sand iris]

Yeah

[sand iris]

Can you screenshot the whole error

I'm 90% sure it's even giving you an error message with a fix !

[peak tendon]

same issue (base)@MBPvonEmanuele Retrieval-based-Voice-Conversion-WebUI-main % source/bin/activate && pip install -U soundfile
zsh: no such file or directory: source/bin/activate

[sand iris]

error: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.microsoft.com/visual-cpp-build-tools/

https://cdn.discordapp.com/emojis/1104640088840876093.webp?size=48&name=clueless&quality=lossless

[peak tendon]

and for the nvidia version i get this : ValueError: mutable default <class 'fairseq.dataclass.configs.CommonConfig'> for field common is not allowed: use default_factory

[sand iris]

it's doing the same thing we fixed before

using the wrong python version

[latent kettle]

Should I install it on my system?

[sand iris]

.

[sand iris]

yes !

[peak tendon]

on my desktop i dont have the persmission to do changing the run.sh

[latent kettle]

And then how to continue?

[peak tendon]

and the other issue where soundfile was not working? i tried the command but no result

[sand iris]

have you downloaded and executed the thing ?

[sand iris]

the pip command ?

That should've done something

[peak tendon]

yes nothing happened

[sand iris]

really not sure what's up with it

it's complaining about soundfile which should be installed

[peak tendon]

ModuleNotFoundError: No module named 'soundfile'

[peak tendon]

i am in my rvc folder

or do i need to use bash

[peak tendon]

source/bin/activate && pip install -U soundfile so this needs to be executed in the rvc directory?

[sand iris]

yeah

[sand iris]

Okay, can you go back to the visual c++ installer ?

Anc screenshot the components that are selected

[peak tendon]

!pip install soundfile does this help?

[latent kettle]

I'm installing it !

[sand iris]

That's for google colab notebooks

or regular notebooks

not related to python code we're running here

[sand iris]

Also, small nitpick, but spaces in folder names tend to cause issues (talking about E:\Artificial Intelligence)

[peak tendon]

with the nvidia version i get now that after running the skript:
ImportError: dlopen(/Applications/RVC1006Nvidia/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <590C66FA-DB62-3794-848C-4DC3F9EA2E70> /Applications/RVC1006Nvidia/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

[rare goblet]

Ayo? @peak tendon level 5 !!!

[latent kettle]

So I have to rename it ?

[peak tendon]

do you know a solution with the nvidia version?

[sand iris]

yeah

unrelated to the error you are having rn though

[latent kettle]

Will it impact other things?

[sand iris]

Apparently it's just broken on m1 macs

I'm reading that the fix was :
source .venv/bin/activate && ARCHFLAGS="-arch arm64" pip install lxml

wait a sec

[peak tendon]

on the NON nvidia version now i got this lol? ModuleNotFoundError: No module named 'librosa'

[sand iris]

won't work

[peak tendon]

it seems like soudnfile was fixed

[sand iris]

Oh FFS what's up with the python dependencies

yeah give up on the nvidia version lol

[foggy spire]

is there any help seeing what to do with the voice settings on the app?

[peak tendon]

but what to do now

[sand iris]

Same thing yet again, !!!
source .venv/bin/activate && pip install -Ur requirements.txt

should fix it once and for all

[foggy spire]

@sand iris check dms pls

[sand iris]

what is it you can't send here ?

[foggy spire]

a picture

[rare goblet]

Ayo? @foggy spire level 1 !!!

[foggy spire]

oh

now i can

[sand iris]

oh yeah lol

[foggy spire]

how do i check if the voice changer works-

[peak tendon]

@sand iris what does that mean

[sand iris]

Great, pick the C++ thing, and look at the right part of the UI

you should only pick

• windows 11 sdk
• MSCV v143 [...]

[sand iris]

#🔍│help-w-okada

[sand iris]

does what mean ?

the command I sent ?

[foggy spire]

is there a tutorial or do i ask btw

[sand iris]

Use the local python environement for RVC, and just reinstall all the packages needed as dependencies

[peak tendon]

execute that in the folder?

[sand iris]

yeah

[latent kettle]

Desktop development C++ ?

[sand iris]

yes!

[peak tendon]

is it ok?

,<3.11 are supported.
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.
(.venv) (base)@MBPvonEmanuele Retrieval-based-Voice-Conversion-WebUI-main %

[sand iris]

https://cdn.discordapp.com/emojis/1026989550628778005.gif?size=48&name=huhcat&quality=lossless

how the hell is it on 3.11

[peak tendon]

this pops up after your new command error: metadata-generation-failed

[peak tendon]

so i delete agian the venv folder and execute ?

[sand iris]

is this the RVC folder with the fixed run.sh ?

[peak tendon]

yes..

the added command line is also with 3.9

i deleted the venv folder again and try running the command agian

[sand iris]

this in a nutshell

It should work

[peak tendon]

i will try to delete venv folder again and execute run.sh

[latent kettle]

What should I install?

[peak tendon]

after running again the run.sh now i got that:
mportError: dlopen(/Users/emanueleg/Documents/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <590C66FA-DB62-3794-848C-4DC3F9EA2E70> /Users/emanuelegemito/Documents/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

what to do now? @sand iris

[sand iris]

Alright

[peak tendon]

at least this was working: source .venv/bin/activate && pip install -Ur requirements.txt

[sand iris]

yeah great

next step is fixing that thing

Which seems to have issues with arm-based macbooks out of the box

[peak tendon]

and the thing is my terminal shows (.venv) at the beginning of every command

[latent kettle]

@sand iris

[peak tendon]

(.venv) (base) emanueleg

[sand iris]

only tick "MSVC v143" and "WINDOWS 11 SDK"

the rest is not needed

[latent kettle]

I'm using windows 10 x64

[sand iris]

yeah it's confusing

but windows 11 sdk is the correct one no matter what

Don't ask me why, but it's like that

¯_(ツ)_/¯

[latent kettle]

It's about 5.86 GB

[sand iris]

Yup

[peak tendon]

and if i move the folder to my desktop it says ModuleNotFoundError: No module named 'librosa'

[sand iris]

source .venv/bin/activate && ARCHFLAGS="-arch arm64" pip install lxml --upgrade --force-reinstall --no-cache-dir

[sand iris]

Oh yeah

moving venvs around is gonna cause issues

[sand iris]

https://cdn.discordapp.com/emojis/739053025125204041.gif?size=48&name=kkkkek&quality=lossless

[latent kettle]

After downloading and installing how can I continue? Should I have to do everything again? Download files etc...?

[sand iris]

just re-run the script that complained about fairseq not being able to build

[distant lily]

where can i get a girl voice model

[sand iris]

should work

[peak tendon]

i executed now that, i moved the folder to my main directory to be safe

[distant lily]

yo

where can i get a girl voice model

[peak tendon]

what does it mean? what do i need to do

[sand iris]

The thing is you want to install it somewhere and just not move it afterwards

if you move it it will die 😆

[peak tendon]

your last comment worked, now again run.sh?

[sand iris]

if you didn't move it then sure

[peak tendon]

again that ModuleNotFoundError: No module named 'librosa'

[sand iris]

Okay, let's make sure everything is okay :

Hopefully this'll do

[peak tendon]

i will not move it anymore

[latent kettle]

pip install -r requirements.txt

?

[sand iris]

👍

[latent kettle]

So I have to download all the stuff again 😰

[sand iris]

1. Choose an install folder. The folder path (/whatever/the/hell/it/might/be) should NOT contain any spaces
2. extract the RVC code there
3. fix run.sh with the thing from earlier
4. execute run.sh
5. hope it works (it should)

[sand iris]

it caches it

should be fine

[glad zealot]

what are they trying to do?

[peak tendon]

what you mean with rvc code here? and what you mean with 1)? i donwloaded it already

[sand iris]

RVC install on macos for insightroam
aicovergen on windows for dr jr

[sand iris]

you got a .zip file right ?

[glad zealot]

ah

[peak tendon]

yeah than i will redownload it right?

[sand iris]

for RVC, from github (not the nvidia one)

[latent kettle]

Does it continue form that step where it stopped?

[sand iris]

if you have the zip already there is no need to

[peak tendon]

i will delete now everything. do you have a link?

yeah i deleted the zip

[sand iris]

yes and no. It tries to redownload everything again unless it finds that it has already downloaded them

just run, you'll see what i mean

[glad zealot]

isnt there a prebuilt mainline version for mac also?

[peak tendon]

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI this link where i click on download zip right?

[sand iris]

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

yes

[sand iris]

nope lol

I mean that's probably better anyways

[glad zealot]

ah i was thinking of voice changer

nvm

[latent kettle]

But it's for 3rd time I'm repeating these steps due to some other errors 😅

[sand iris]

since this means m1/m2/m3 users can get something that work

[sand iris]

it's erroring again ?

[latent kettle]

No now I'm downloading MSVC and Windows 11 SDK

[latent kettle]

Before that I got some common errors such as pip version is old etc.. and some times it lost internet connection 😤

[sand iris]

the pip is old stuff is like only a warning

you can ignore it

the important part was the c++ build tools

which once installed will fix anything

heck, you can even ask @ lubert_ , I made him do the same thing and it fixed it for him

[latent kettle]

But i have updated it ! As it shows a command to update it

[peak tendon]

@sand iris i downloaded from your link, changed the line in the run.sh and now i am executing it - correct?

[sand iris]

You made sure the folders leading to the run script didn't contain any spaces right ?

[peak tendon]

if i copy the path it shows that: /Users/emanueleg/Retrieval-based-Voice-Conversion-WebUI-main

seems correct, right?

[sand iris]

yeah

[latent kettle]

I'm unable to rename the folder should i close CMD?

[sand iris]

yeah, unless there's something running on it

[latent kettle]

No there is nothing which is currently running! I'm just in current directory which I want to rename

[peak tendon]

I am getting this now:
ImportError: dlopen(/Users/emanueleg/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <590C66FA-DB62-3794-848C-4DC3F9EA2E70> /Users/emanuelegemito/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

[sand iris]

okay great

I was expecting that one this time

should be the last issue

source .venv/bin/activate && ARCHFLAGS="-arch arm64" pip install lxml --upgrade --force-reinstall --no-cache-dir

[peak tendon]

into the folder right?

[sand iris]

yup!

[peak tendon]

ok, its downlaoding something. lets wait..

ok done

and now?

[latent kettle]

Should I close cmd ?

[proud elbow]

close CMD or change to another directory

[peak tendon]

its finishd now, should i run again run.sh?

[sand iris]

if it's not running anything important yeah

[sand iris]

yes

[latent kettle]

Okay but still I'm worried. How can I continue? Because I don't want to download that stuff again and again

[sand iris]

it's on your disk

it'll resume

[peak tendon]

i get that again:
ImportError: dlopen(/Users/emanueleg/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <0083EE08-E132-3DF7-B98B-6824B36DAF1D> /Users/emanuelegemito/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: tried: '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/opt/homebrew/lib/libxslt.1.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/lib/libxslt.1.dylib' (no such file)

[sand iris]

oh ffs

[latent kettle]

So I have to get in same directory and it will continue?

[peak tendon]

does it make sense to download the missing files manually?

[sand iris]

yeah

[peak tendon]

or maybe should we change to phyton 3.8?

[proud elbow]

then you'd better close CMD, rename the folder, than open CMD again to the recently renamed folder

[peak tendon]

what do yo say

[latent kettle]

Can you tell me where these files are getting saved? And how i can I change the path! Because my c drive have only 60GB left

[sand iris]

Not sure how you did it, but it might be within your user's python installation

so no big deal, it'll resume

[peak tendon]

so no possibilities on my end? 😦

[rare goblet]

Ayo? @peak tendon level 6 !!!

[latent kettle]

how i can I change the path! Because my C drive have only 60GB left

[sand iris]

There's a lot of different ways python stuff can be installed

I'm honestly not sure

[peak tendon]

for me no more chance to get that working?

[sand iris]

c:\users\admin\appdata\local\pip\

[latent kettle]

😵‍💫

[sand iris]

I'm looking it up, but I really don't understand why or how it's doing that

[peak tendon]

it gets immediately to this error message after trying to execute run.sh again

[latent kettle]

So what that directory is doing which I have selected? E:\ Artificial_Intelligence\AiCoverGen

[sand iris]

It's installing to the global python installation apparently

So it's on C:\ no matter what

[latent kettle]

But I don't want to install it in C: 😐

[rare goblet]

Ayo? @latent kettle level 11 !!!

[sand iris]

gonna have to dip soon, installing in other places is a little more manual

[latent kettle]

😅

[peak tendon]

any more ideas?

[sand iris]

¯_(ツ)_/¯

[peak tendon]

if we download the needed files manually?

or using 3.8 phyton?

[sand iris]

I don't have macos to test on unfortunately, the closest I have is linux, which installs just fine with the same instructions

tried both 3.9 and 3.10 and they both worked

https://cdn.discordapp.com/emojis/924763610197147678.webp?size=48&name=hmmm&quality=lossless

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/804

[ivory surge]

Can someone help me with aesthetic_rvc_inference_hf? I'm putting different models in the inferencing voice but all the audios end up with the voice from the first model I downloaded there

[peak tendon]

and now..

[fallow edge]

hey is it still possible to use rvc like how you could a few months ago on the website or is it only via code now

[peak tendon]

it says always: A new release of pip is available: 23.0.1 -> 23.3.1

could be this the issue?

[peak tendon]

here it says: Uninstall all packages including cpython, then run "run.sh" which will install all packages you need and run automatically.

how this can be done?

[sand iris]

Unsure

[glad zealot]

what are they gonna use it for? training?

[peak tendon]

shit..

[peak tendon]

i deleted every python installation and now i got a different error executing run.sh:
ImportError: dlopen(/Users/emanueleg/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so, 0x0002): Library not loaded: @rpath/libxslt.1.dylib
Referenced from: <590C66FA-DB62-3794-848C-4DC3F9EA2E70> /Users/emanuelegemito/Retrieval-based-Voice-Conversion-WebUI-main/.venv/lib/python3.9/site-packages/lxml/etree.cpython-39-darwin.so
Reason: no LC_RPATH's found
do you know what this is

[merry sand]

I tried training a voice and I can't get the path nor model files. Just the rvclogs.

[glad zealot]

find a folder called weights

it should be there

[merry sand]

That's not on my Drive file list.

[glad zealot]

check colab not drive

on the left side there should be a file list cehck it there

[merry sand]

Do I also have to load the model I trained to find it?

[ashen bison]

hey y'all i downloaded a oice model with a g and d file, how do i get the .index file using applio?

[glad zealot]

the g and d are not the usable models

[ashen bison]

oh dang

[glad zealot]

wut

[glad zealot]

check a folder called weigths

somewhere...

[brisk pebble]

Guys

[merry sand]

You know, to find the file in the collab.

[brisk pebble]

No sound output can anyone help me?

[glad zealot]

you are using rvc disconencted right?

[glad zealot]

#🔍│help-w-okada

[merry sand]

Yes.

[brisk pebble]

Ok

[glad zealot]

check the logs and if theres a weigths folder inside of it

[merry sand]

And I ended the runtime after training the model, but I saw nothing but RVClogs.zip.

[glad zealot]

ahh if you ended you prob need to re train

since you need to get the model before stopping i

t

[merry sand]

Oh

[merry sand]

BTW, in GC, I was wondering what this is.

And after the model is done training in the collab, I have to look for it in the weights in the collab?

[proper shale]

This allows you to see how training is going

[proper shale]

You run the "Export Model" step

[merry sand]

Ah

[proper shale]

And it'll save on Drive

[proper shale]

-overtrain

[azure marsh]

Hey, @merry sand!

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

How do I know if my model is overtaining?

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

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

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[proper shale]

The Guide will help you identify overtraining

[merry sand]

Although all I see is rvclogs.zip.

[proper shale]

If you're training rn, it won't run that step until you stop the current training step

[merry sand]

Oh. So after it's done training, do I have to also run "Export Model from Notebook to Drive' to get the model to Drive?

[proper shale]

Yeah.

I don't recommend training for over 2 hours though, so once it reaches a save epoch, wait until the next epoch, stop the training step and export the model

[charred canopy]

can someone help me im trying to look for popular models but its only showing me the most recent uploads in ⁠🎧┋voice-models

[merry sand]

It's a short part with just a 2 minute, 40 second audio thingy. So it shouldn't take that long. And I'm trying to get it to reach 300 Epochs in total. But after that, I esport it, right? So I wouldn't have to just see rvclogs.zip all the time?

[proper shale]

Yeah you just export it

It should reach 300 epochs in around an hour or so

[proper shale]

Right now there isn't a way to search for popular models

Only weights.gg, I guess

[charred canopy]

what is weights gg

[glad zealot]

its a website with all of the models in this server+more ig

[proper shale]

^

[charred canopy]

even on the website the most liked ones sound so bad its insane

something must be wrong

[glad zealot]

the test audio they have sucks

[merry sand]

Or less, because it's short.

And it's 84 voice clips of a Kung Fu Chaos character apparently. I found a way to export audio for OG Xbox games.

[rare goblet]

Ayo? @merry sand level 7 !!!

[glad zealot]

noice

300 is too much cause pretty sure colab is gonna close before it even reaches taht

[merry sand]

It reached. :)
Also I exported and trained the index, so I think I got it now.

[glad zealot]

oh damn

[dire wagon]

@rapid grail

[rare goblet]

Ayo? @dire wagon level 1 !!!

[latent kettle]

Can I use AIcovergen without nvidia GPU?

[proper shale]

You can, but it'll be slow

[latent kettle]

How?

It is showing me an error

[proper shale]

I mean, I haven't tried it so I wouldn't know but it probably does work

[proper shale]

Hmmm is there any error message or something

[latent kettle]

Yes ! It's like "your system doesn't have nvidia drivers

And I'm waiting for a long time for it's installation but at last I got this

[proper shale]

oh ur using locally?

yeah don't think it'll work then

[latent kettle]

Why are you saying this 😩

I have wasted my 1-2 hours for this 😐

[proper shale]

Don't wanna make you lose more time

[latent kettle]

So I think it's useless 🙄

[merry sand]

Hehe, ye. Thanks for the help though, I forgot about the export and index training parts, as I haven't trained an RVC model for months.

[brittle wing]

how do i train with rvc disconnected?

[proper shale]

there's a guide in #1159513888199540817, just scroll down a bit

[brittle wing]

how should i setup my data set, like should it all be one file or seperate of all my acapellas

[latent kettle]

You have to prepare a dataset

[latent kettle]

As your wish. Both results same

[brittle wing]

ok thx

[proper shale]

One file is generally better

[latent kettle]

I think yes! But there is no huge difference!

[proper shale]

Actually makes a difference in speed - 1,2k 1-3 sec files take 1 min 30 secs to train one epoch, meaning more files hurt performance

[latent kettle]

Yeh that's true! But i think output remains same in both cases

[young halo]

Ok, I think my model is done training, what do I have to do now?

[proper shale]

have you exported the model?

if so u can exit the page

then just zip pth and index, upload to huggingface and use it

[young halo]

How can I do that? I only saw that it was donde

*done

[proper shale]

There's an export model step

[young halo]

Is it on the RVC disconnected?

[proper shale]

yep

[young halo]

Ok let me see

Is that one?

[proper shale]

yup

[abstract flame]

yooo, how do i make the ai things?

[proper shale]

#1159513888199540817

[abstract flame]

yeah but i don't want to use my gpu

[young halo]

Ok, I think it's done, what do I do next? Sorry to bother you, this is my very first time making a model

[latent kettle]

Now go to your Google drive

And you will see a folder RVCDISCONNECTED

[young halo]

One of those right?

[latent kettle]

Open it and then you will see 2 files one is (.pth) and another one is (.index)

[proper shale]

then just use the colabs

-colab

[azure marsh]

All working colabs / spaces ☁️

• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• 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

Credits to Faze Masta and Antasma for compiling these links.

[abstract flame]

wich one of them

[latent kettle]

Open these folders and find the files

[young halo]

Ok, I see these

[rare goblet]

Ayo? @young halo level 3 !!!

[proper shale]

what do you wanna do, covers or train models?

[abstract flame]

do covers

[proper shale]

download the youngji.pth

[abstract flame]

do i use this one without using my gpu ?
https://huggingface.co/spaces/TheStinger/Ilaria_RVC

[proper shale]

then use AICoverGen

[proper shale]

or this one yeah, theres a guide for it unlike AICG

[young halo]

Done

[abstract flame]

bet, thanks

[latent kettle]

Okay download first and last one..

[rare goblet]

Ayo? @abstract flame level 1 !!!

[young halo]

Done

[latent kettle]

Same for index

[proper shale]

you're welcome 🤗, good luck on the covers!

[young halo]

Got u

[latent kettle]

Now go to download folder in your computer

[young halo]

There?

[latent kettle]

Yes !

A question* do you have winrar? Or any archiver ?

[young halo]

Yes I do!

[latent kettle]

Which one? Winrar?

[young halo]

Yes I think so, Is that the one that looks like books?

[proper shale]

now zip the index with the pth and upload to huggingface

[latent kettle]

Yes absolutely !

[young halo]

Alright!

[latent kettle]

Now select both files

[young halo]

Got u

[latent kettle]

By pressing Ctrl key and select files by mouse left key

[young halo]

Yes!

[latent kettle]

Okay.. now hold Ctrl key and right click on it

[young halo]

Done

[latent kettle]

And choose add to archiver!

[young halo]

Do I select zip carpet?

*Folder

[proper shale]

nah just zip the two files

[latent kettle]

Let me see! Please send me a screenshot

[proper shale]

i understand a bit of Spanish so maybe i can understand what the words mean lol

[proper shale]

yeah that

[young halo]

Alright, do I change the name?

[latent kettle]

Select zip

[young halo]

Ok

[young halo]

Done!

Do I have to change the name or just leave it like that?

[latent kettle]

Rename it according to you. Don't include space " "

[young halo]

Ok

Done

[proper shale]

Now just upload to huggingface

[young halo]

It's called youngji.zip

[proper shale]

-huggingface

[azure marsh]

huggingface.co upload Guides

English guide

Link: https://rentry.org/fdg_guide_new
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

[young halo]

Ok!

[proper shale]

there's the guide if u need it

[young halo]

Is this right?

[young halo]

Thank you!

[latent kettle]

Now you will see a zip file

[proper shale]

yea

[proper shale]

np :)

[latent kettle]

It can also done by Google drive links?

[rotund zephyr]

i said this on the other chat but like my voice cuts off, for example i say "Hello my cat has white fur" and it'll only output the, "has white fur" part, i think its my internet but there could be other causes

[young halo]

Mmmm, this is weird, and it's not giving the option of adding a file

[proper shale]

go to your profile and click on the model page

should have an option to add stuff

[proper shale]

Increase your chunks to like 440, and decrease s thresh

if it's increased

[young halo]

Here?

[proper shale]

files and versions

[latent kettle]

Ai covergen offers upload models option so it will be more easy to upload models for local system

[rotund zephyr]

i have s thresh at zero and sup2 activated

[proper shale]

I'm aware - but having ur model in Huggingface is just better

Than Google Drive

[young halo]

And then add file right?

[proper shale]

yup

[latent kettle]

Could you explain why ? Because I really want to know!

[proper shale]

you have a direct zip download link to your model

it works everywhere basically

[young halo]

Done

Now do I press commit?

[proper shale]

Now just commit changes and off you go

[rotund zephyr]

also my advanced settings has silencefront off and rvc at high idk if that does anything

[proper shale]

Maybe keep rvc quality low?

Also, increase gain in a bit, maybe the start of a sentence isnt that loud

[rotund zephyr]

which gain? the one in the audio section? or the one next to the model

[young halo]

Ok it's uploaded now!

[latent kettle]

Now go to colab to use it

-colab

[azure marsh]

All working colabs / spaces ☁️

• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• 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

Credits to Faze Masta and Antasma for compiling these links.

[young halo]

Alright! But how do i get the link?

[latent kettle]

These are some colab Notebooks

[young halo]

The link to the model

[brittle wing]

how many epoch for like a 9 min dataset

[latent kettle]

Depends on your data! I prefer 300-500

Or more than it

[brittle wing]

ok i had it at 500 i just know its gonna take forever

[latent kettle]

ARe you training it on local hardware or on colab?

[brittle wing]

on colab

[latent kettle]

So just set it on 10 firstly, to estimate the time of training

[brittle wing]

ok

[latent kettle]

Check the average training time for per epoch!

[young halo]

My model works just perfect, thank you so much for your help, it wouldn't be possible Without You!!

[latent kettle]

Congratulations 🎉 and best of luck for future covers

[latent kettle]

Done with 10 epochs?

[young halo]

I appeciate it!

[rare goblet]

Ayo? @young halo level 4 !!!

[brittle wing]

yea

[latent kettle]

What about timing for per epoch?

[brittle wing]

wdym?

[rare goblet]

Ayo? @brittle wing level 4 !!!

[latent kettle]

While training you can see there is a time to train per epoch!

[brittle wing]

23.673000

[latent kettle]

For example you have trained 10 epochs so how much time per epoch takes to train!

[latent kettle]

1 - 10 all epochs take same time?

[brittle wing]

no they take like either a second more or a second less but theyr all around 22,23,24

[latent kettle]

Let me calculate. Because colab offers only 4 hours of Free GPU resources

[latent kettle]

Firstly set epochs to 400-450 and after that you can continue training if you want want!

[brittle wing]

oki thx

[latent kettle]

Yes but remember you have only 4 hours! After that you have to wait for 22-24 hours

Or you have to use an alternative account

[rare goblet]

Ayo? @latent kettle level 12 !!!

[brittle wing]

oki

[molten pecan]

but you can keep training the model from where you left, right?

[latent kettle]

Yeh, absolutely

[molten pecan]

phew, thank god

[latent kettle]

What happened?

[molten pecan]

nothing, just asking

[latent kettle]

Yes We can continue! I have already mentioned above

[slate pawn]

Anyone knows how to fix the issue of having the index file not working?

[ripe whale]

anyone know if there are any colabs/spaces able to do model fusion?

[fiery torrent]

How do I make a model? Is there like a guide?

[azure minnow]

WHAT THE FUCK.

[proper shale]

#1159513888199540817

[hot lagoon]

have there been modle with just 1 epoch

[crimson wraith]

n00b question: if using the colab is the easiest way and doesnt require a decent PC, then why do people still get graphics cards and whatnot to do it locally? Just trying to figure out if I'm missing something and am correct in that I dont need to go get a 3080 or anything like that.. I don't mind paying the $10/mo for the colab pro if that helps with ease and speed of training.

[molten pecan]

what's the problem?

[molten pecan]

buying a GPU allows you to train even faster than colab pro. also you can run games and AI tools much faster

[meager terrace]

Any possibility to get covers not capping at 15Hz?

Used illaria RVC

@brave garnet

[rare goblet]

Ayo? @meager terrace level 10 !!!

[violet heron]

Demo should be a sample of a cover using the model, DO NOT post instrumentals.

[finite galleon]

Try this https://colab.research.google.com/drive/1X8YR4Ruv7zzY8YAMPTfC7hkxqT_d4Q5d

I havent tested if it has a cutoff or not sooo

Just test it ig

[meager terrace]

But how do you do it yourself?

You get no cutoff?

[finite galleon]

?

[finite galleon]

I never looked an inferred file spectogram lol

[meager terrace]

Ok, spsb

❤️

[brittle wing]

how do i fix Your runtime has been disconnected due to executing code that is disallowed in our free of charge tier. ?

its just does that when im gonna use the T4

can i use TPU

[violet heron]

Which colab?

[young halo]

Can you please send me a collab for making covers?

[violet heron]

-colab first one

[azure marsh]

Suggestion for @young halo

All working colabs / spaces ☁️

• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• 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

Credits to Faze Masta and Antasma for compiling these links.

[brittle wing]

google colab

[violet heron]

No i meant which colab/notebook

Send a link

[brittle wing]

https://github.com/w-okada/voice-changer/blob/master/Realtime_Voice_Changer_on_Colab.ipynb

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

this one @violet heron

[violet heron]

Outdated

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

Use this ON A ALT

[celest ether]

i keep getting a message saying “Error None”

[violet heron]

Using….

[celest ether]

https://huggingface.co/spaces/r3gm/RVC_HF

[rare goblet]

Ayo? @celest ether level 1 !!!

[celest ether]

this site

[violet heron]

Did you duplicate the space?

[tame mica]

hina this u?

[celest ether]

🤨🤨🤨

[celest ether]

how what

[graceful owl]

Anybody got experience with the new RVC update and tell me if its worth the headache

[violet heron]

3 dots at the top bar, duplicate this space

Need an account

[celest ether]

oh okay

yay

[azure minnow]

It won't let me attach my image or my voice sample, like at all

[molten pecan]

Dang that's weird. Ask @proper shale

[tame mica]

prob not leveled up yet

normal

[molten pecan]

ah

[celest ether]

i did but i still have the same problem

[violet heron]

Show a full screenshot

[violet heron]

Does the audio file name have any spaces?

[celest ether]

no :'0

idk why but it worked now

[rare goblet]

Ayo? @celest ether level 2 !!!

[celest ether]

oop

thank you for your help C:

[modern surge]

Failed to load library libonnxrunt

[brittle wing]

this is the new one?

[violet heron]

Yep

[ashen gazelle]

I keep getting this error: "RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory" any suggestions?

[brittle wing]

help

i want to upload my model

i followed the instructions

google drive link

[glad zealot]

nyo...

https://cdn.discordapp.com/emojis/650668104828977155.webp?size=48&name=Sweat&quality=lossless

[glad zealot]

can you send the drive link?

[brittle wing]

nvm i can just upload via the web ui

thanks tho

CAN*

[glad zealot]

i think i need to update the drive uploded too probably

[ashen gazelle]

[Voice Changer] use torch contentvec [ONNXRuntimeError] : 7 : INVALID_PROTOBUF : Load model from pretrain/content_vec_500.onnx failed:Protobuf parsing failed.
[Voice Changer] exception! loading embedder PytorchStreamReader failed reading zip archive: failed finding central directory cpu

Any way to fix this?

would reinstalling all directories help?

[rare goblet]

Ayo? @ashen gazelle level 1 !!!

[brittle wing]

what does Index do??

[ashen gazelle]

Sort of calibrate the ai matching to your voice

Most don't really notice a difference though

[molten pecan]

.index is a file that contains the features from the dataset, like accent, speech manner, etc. from an RVC voice model. This file is not a requierement for a model to function, but it makes it sound much better

[brittle wing]

so what if i do 1 index?

or 0.5?

[molten pecan]

are you training a model?

[brittle wing]

no

[glad zealot]

accent basically or intonation

[brittle wing]

im just voice changing my voice

[glad zealot]

you can just leave index at 0 then

[brittle wing]

i just got curious what it does

[glad zealot]

or dont even upload index

[white bough]

Is anyone else having GPU problems while using ardha's colab?

[young halo]

Does someone know how to get more GPU?

[molten pecan]

in Colab or Hugging Face?

[young halo]

Colab

[molten pecan]

you could have googled it 😭

[young halo]

So i have to buy it?

😭

[molten pecan]

mhm

[glad zealot]

or alt but ye

[young halo]

Or is it going to refill by itself?

[violet heron]

No

[violet heron]

Refills every 4 days iirc

You can always use alts

[young halo]

How can I use those?

[molten pecan]

alts are alternative accounts

[young halo]

Ooooo

[molten pecan]

alt google accs

[young halo]

I see

Thank you!

[molten pecan]

yw hf!

[plush imp]

I wanna make an ai cover, but how do I make it with it saying different lyrics? Like putting gyatt, and skibidi in it?

[winged fulcrum]

rq question because i forgot but how many epochs for about 15 minutes of training data

[winged fulcrum]

best thing to do is just record those extra lyrics and insert them where they gotta go

[thin stump]

Trial and error with inputting lyrics in Synth V > Running Denoise on UVR 5 > then generating it with the AI model

Or just sing it yourself with Okada

[dire wagon]

Could not find TensorRT
Please install tensorboardX: pip install tensorboardX

is there a way to fix this ?

[rare goblet]

Ayo? @dire wagon level 2 !!!

[fiery torrent]

how do i know when to stop training?

[acoustic garnet]

Tensorboard graphs

or just overtraining like 400 epoch then checking them model though inferencing

[molten pecan]

@fiery torrent

[fiery torrent]

i saw

ive never used tensorboard before

how do i use it

[acoustic garnet]

local or colab?

[rare goblet]

Ayo? @acoustic garnet level 4 !!!

[livid harbor]

One message removed from a suspended account.

One message removed from a suspended account.

[acoustic garnet]

thats just a really shit headset?

'echo cancellation' is probably the thing you need to cut off such things.

[chilly peak]

if i'm training a model and some of the sound files it trains on have the character doing grunts and loud breathing (but all files have at least some speaking), would this affect the AI's performance for copying the voice ?

[copper pebble]

probably

[chilly peak]

good to know thanks!

[azure obsidian]

can i use my model to TTS? in ilaria RVC

[proven hill]

yes

[azure obsidian]

how i can

i got my model but idk how to use it to TTS

[proven hill]

chose a voice in the tts based on the language if the tone is masculine or feminine

[azure obsidian]

my own model

[rare goblet]

Ayo? @azure obsidian level 1 !!!

[proven hill]

yes, same way

you downloaded it?

[proven hill]

yea then follow the steps im saying

[azure obsidian]

ok

[proven hill]

input the text here

press speak after

immediately after it will look like this

now you can convert

[azure obsidian]

voice looks weird

its not same as the model i have

[golden karma]

Hey guys juts wanna ask, is there any way to separated 2 vocal that overlapping via UVR? i mean when 2 singer singing together? or maybe any suggestion?

[rare goblet]

Ayo? @golden karma level 1 !!!

[lament eagle]

What is pitch_guidance

is it true that when training model with pitch_guidance off it'd make it better

like singing model at least

[lament eagle]

or is it depends on dataset?

[bronze crane]

why does the vb cable output show up as input for me?

[sand iris]

expected

[bronze crane]

?

[sand iris]

That's just how it works

[fiery torrent]

how long should training the feature index take?

bc its stuck on generating adding file

[chilly peak]

i think it took a couple seconds for me

[brittle wing]

anyone knows how to fix on mac?

Traceback (most recent call last):
File "infer-web.py", line 27, in <module>
import gradio as gr
File "/opt/homebrew/lib/python3.8/site-packages/gradio/init.py", line 3, in <module>
import gradio.components as components
File "/opt/homebrew/lib/python3.8/site-packages/gradio/components.py", line 32, in <module>
from fastapi import UploadFile
File "/opt/homebrew/lib/python3.8/site-packages/fastapi/init.py", line 7, in <module>
from .applications import FastAPI as FastAPI
File "/opt/homebrew/lib/python3.8/site-packages/fastapi/applications.py", line 15, in <module>
from fastapi import routing
File "/opt/homebrew/lib/python3.8/site-packages/fastapi/routing.py", line 22, in <module>
from fastapi import params
File "/opt/homebrew/lib/python3.8/site-packages/fastapi/params.py", line 4, in <module>
from pydantic.fields import FieldInfo, Undefined
ImportError: cannot import name 'Undefined' from 'pydantic.fields' (/opt/homebrew/lib/python3.8/site-packages/pydantic/fields.py)

[hallow widget]

[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
How to fix

[brittle jay]

Hey, why is text-to-speech getting errors on weights?

it said error in red where "cancelled" says in the pic

[rapid badge]

Hi … I already run Manio RVC and Apollio RVC locally in windows. But im mac its always having issues. Please suggest any other versions to run locally in mac . TIA.

[meager terrace]

Do you know another cover colab but not capping at 15Hz?

[proven hill]

for training?

[meager terrace]

For making covers

Caus with the Ilaria covers my songs are capping at 15Hz

[proven hill]

i dont use colabs

[proven hill]

thats the technology, wont change much, youre probably using rmvpe

[meager terrace]

O so the reason is the RVC training not the cover tool we use?

Right?

hmmmm

[proven hill]

should be the tecnology behind RVC

since especially the output for example is mono