✨│ai-help

[viscid moss]

I mean, u need to download the version for ur PC specs (GPU)

[nocturne patrol]

I have a 1070

ohhh

[viscid moss]

hmmm

[nocturne patrol]

i have nivida gtx 1070

my pc is not the best bc its on a budget but ill try

[viscid moss]

It will work still

[nocturne patrol]

i cant enter the sight

im so cooked

[viscid moss]

https://huggingface.co/Shadicti/deiteris-Fork/blob/main/voice-changer-windows-nvidia-b2332.zip

download this first

then extract

[nocturne patrol]

alrr

[twilit forge]

should i just stop the training now or should i see if it goes back down

[viscid moss]

I think it won't go down

so ya

[twilit forge]

bet

[nocturne patrol]

IT STILL DOESNT WORK

do i need to download the vb audio cable bc i uninstalled it

even tho it still shows in the thing..

it lets me do it on server but not client and idek what im doing anymore

[elder coral]

What cloud

[nocturne patrol]

?

[elder coral]

wHAT SHOULD I TRAIN ON COLAB

[eternal aurora]

where should i add my token

[elder coral]

I'm gonna train a robot voice

[fading lodge]

can i use UVC to just remove bg noise on audios and stuff

i got a voicemodel someones making for me but he needs more dataset and im wondering if that works

[viscid moss]

hmmmm can u send a screenshot of the CMD? Is there any error?

[viscid moss]

a voice model

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[viscid moss]

yep

it depends on what type of bg noise tho

[fading lodge]

not much

its mainly just like bits of music or backround noise on my end when someone messages me

[viscid moss]

that could be removed with UVR

[fading lodge]

alr

[viscid moss]

prob with MelBand Roformer | Vocals FV4 by Gabox

[fading lodge]

i was using MDX23C

[knotty moth]

this is not 2023/2024 anymore

[fading lodge]

Thats what it said on the document

[nocturne patrol]

it still doesn’t let me put my input in

[simple ore]

replace the word 'token'

[dense vale]

I'm noticing a weird issue with my fresh install of the deiteris fork of w-okada/wokada; I'm on Windows, and I grabbed the voice-changer-windows-amd64-cuda.zip.001 and 002 from https://github.com/deiteris/voice-changer/releases. I'm just following the install guidance at https://github.com/deiteris/voice-changer.

Whenever I attempt to launch by double-clicking MMVCServerSIO.exe, I get a strange crash.

• I double-click MMVCServerSIO.exe
• Terminal opens up
• Lines are written stating I have Python 3.12.7, noting the Voice changer version as b2332 NVIDIA-CUDA, loading weights, and noting all items are downloaded with all weights loaded.
• One final line is written "[main] protocol: HTTP"
• About a second passes
• Loading icon / spinning ring appears on mouse cursor briefly
• Terminal crashes

In effect, the terminal window that opens when double-clicking MMVCServerSIO.exe seems to crash after the "[main] protocol: HTTP" line.

I can see that all of this is suitably being written to vcclient.log, given the file's updating with a new Date modified date/time whenever the terminal crashes... But there's no error recorded. When I open vcclient.log, the final entry is always just "[main] protocol: HTTP"

I'm not exactly sure what I may be doing wrong here; I should be on a stable GPU driver for my 40-series card, and I have no pending Windows updates. BIOS for my MSI motherboard seems to be fine. I'm not having issues running or launching other AI-related local programs, just the deiteris fork of w-okada.

This is additionally extra weird, because I was using w-okada perfectly fine about six months ago in October 2024. I can't tell what might have changed between then and now, if anything.

I just keep seeing the MMVCServerSIO.exe terminal crash on that "[main] protocol: HTTP" line.

It's very strange.

===============
IMPORTANT EDIT

Unplug your monitor from your motherboard. It fixes this. I don't know why.

[modern swallow]

I use Intel(R) Core(TM) i5-8265U CPU can it run rvc stably on discord

[hallow thistle]

You run RVC or W-Okada the realtime voice changer with that CPU? That doesn't sound right.

You got any GPU?

[supple salmon]

Should I use RVC o W-Okada? :o

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[hallow thistle]

That's a bad question. RVC is regular voice changer, while W-Okada is "realtime" voice changer. Both uses RVC voice model to inference, the difference is how they supposed to work.

[modern swallow]

i have but it is intel R UHD graphics 620

[hallow thistle]

That's an integrated GPU, and even worse.

[modern swallow]

yeh ☠️

so what should i do

[hallow thistle]

What should you do? That's one of the most asked questions I've ever seen. Running W-Okada with only CPU isn't so recommended. You can try W-Okada on Kaggle instead, but you'll have to register with your phone number on this one, and ngrok account needed for token. https://www.kaggle.com/code/suneku/voice-changer-public

[modern swallow]

ok =()

[supple salmon]

Terribly sorry for the bad question, though thank you for replying!

I just tried RVC and it is just freezing upon launching the realtime gui

so I'll try wokada

[hallow thistle]

What is your PC GPU?

[supple salmon]

4070ti!

[hallow thistle]

Download and use this better W-Okada. https://rentry.co/ForkVoiceChangerGuide#download-for-nvidia-gpu-on-windows

[supple salmon]

Oh thank you!

Perhaps a bad question again, but these forks are just like distros for linux, right? :o

[dense vale]

This link actually also helped me diag another issue

[dense vale]

With this up here

It's very handy

Specifically in my case, I've tried the... admittedly unorthodox play of attempting to use the 5000-series-specific download, which is mentioned on that Rentry

that version of W-Okada and its MMVCServerSIO.exe will mention process exit codes / error codes

I don't know why that functionality isn't in the main version or the deiteris fork.

[dense vale]

But still, I now know that this behavior up here's happening in my case during the HTTP launch due to some form of access violation

good ol' 0xc0000005

[hallow thistle]

I was about to reply to your message, but the way you word your message is kinda confusing.

[dense vale]

Yeah that's on me

The gist is that running MMVCServerSIO fails at the [main] protocol: HTTP line

I can now see that it's specifically failing / exiting process with code 0xc0000005

This can mean a whole host of potential points of failure varying from RAM issues to CPU issues to motherboard issues to... really many, many things.

In my case, it may also indicate some level of failure in my intel CPU which, to be frank, isn't... unexpected for my gen of i7.

[hallow thistle]

[dense vale]

The issue mentioned up here likely isn't a W-Okada thing, it's probably local to my machine and may be indicative of CPU failure.

[knotty moth]

not surprising that it refuses to run on non-50 series gpus

[nocturne patrol]

?

[knotty moth]

still the virtual cable problem?

tell me if u need to post screenshot

[nocturne patrol]

i didn’t download that

do i need to?

[hallow thistle]

Crazy. Download Virtual Audio Cable lite now. https://software.muzychenko.net/freeware/vac470lite.zip

[nocturne patrol]

okay

boom downloaded

still did not work😞

it keeps saying this

[round hearth]

so i have an Nvidia 1080. i know, super outdated, but it functions for now. i just dunno what the hot new model is nowadays. so i wanted to ask what the hottest new is

[nocturne patrol]

i have a 1070 and they said use this

[hallow thistle]

There's Applio.

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[round hearth]

i'll look at that Applio doc rn

ty

says 2k but i know 1080 is pretty comparable. do you think that's good enough?

[knotty moth]

check the console window for actual error messages (the black background one)

[nocturne patrol]

alr

[hallow thistle]

Any NVIDIA GeForce GTX 10/16 GPU is fine for basic AI program. If you wanna "train a voice model", NVIDIA GeForce RTX 20 series GPU or up is definitely recommended for that task.

[nocturne patrol]

it said switched to cpu

and model slot is not found -1

[round hearth]

noted. I'll probably try and suffer with the slow. buuuuuut i do appreciate you pointing me in the right direction. plus for letting me know which channel was right now. last time i was here there were WAAAAAAAAY more. so TY for all the help. i'll def be back in the future of Mangio's current training while i do HW goes poorly and i need some help with figuring out Applio

[hallow thistle]

You're welcome. Mangio as an RVC program is outdated, as there's no more development going on.

[knotty moth]

mangio is abandoned indefinitely

[sharp lava]

Hi I’ve been using w-okada voice changer, and i rebooted it recently and now every model sounds very robotic, even the base models, how could this have happend?

[deft pewter]

someone help me. im using w-okada normal one last ver of cuda not forked. this never used to happen basically im using the voice changer and yk how theres a terminal running aka the .bat file for running the voice changer basically it terminated itself and the w-okada voice changer didnt close but everything stopped working now when i keep opening it just keeps terminating itself without even me even getting the change to use a model yet

[viral bobcat]

if you wanna use w-okada on a game like roblox, what input and output do you use for the voice changer and roblox? im mac, so im using blackhole

[modern swallow]

how can i use rvc on kaggle

[round hearth]

hi guys. just wondering. which of the UVR models is the best for you guys?

[simple kayak]

hi, wi want to ask. why my voice is not change

[hallow thistle]

The original version of W-Okada is outdated, and very buggy at the moment. What is your PC GPU?

[hallow thistle]

There are two Macs: Intel Mac and Apple Silicon (M CPU) Mac. Running W-Okada locally on any of Mac won't be that good, but I think it's possible to run it from Kaggle.

[simple kayak]

guys i need help, why does my cmd show pipeline not initialized. how to fix that

[hallow thistle]

RVC or W-Okada?

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[hazy belfry]

Hi everyone, I’m looking for an expert who can build a real-time deepfake video and voice system for live Zoom/Teams meetings.

The goal:

• Live avatar (my face) + live AI voice cloning (my voice) during meetings.
• It needs to look natural, respond to conversations, and work with remote laptops.

If you’ve done something like this (or close), please DM me!

This is a serious paid project. Thank you!

Looking for someone who can start immediately or has demos to show. Open to custom solutions or existing tools integration.

[hallow thistle]

This is not where you promote your job application.

[modern swallow]

Uhmm w-okada

[simple kayak]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[hallow thistle]

Ah, it's such a mess when there are many people in #✨│ai-help looking for W-Okada at the same time.

[hazy belfry]

Who’s w-okada??

[hallow thistle]

W-Okada is not a person; it's an AI program where it changes your voice in realtime.

[hallow thistle]

Are you using this one? https://www.kaggle.com/code/suneku/voice-changer-public

[modern swallow]

ye bruh

but i don't know how to use it =v

[hazy belfry]

Hi everyone,

I’m hiring for a paid project: I need a live deepfake system (my face + my voice) for real-time Zoom/Teams calls.

Goal:

• Live avatar of my real face during meetings
• Live AI voice cloning (my voice)
• Must work smoothly with remote laptops and screen sharing

This is a real paying opportunity. DM me if you’re confident you can deliver.

Thanks!

[hallow thistle]

W-Okada not changing your voice can happen when you didn't press green start button on UI, or you just didn't set something right. What is your PC GPU? Did you follow any tutorial video on YouTube before this?

[hallow thistle]

Please stop doing this.

[hazy belfry]

But i need help

It’s called ai support

[craggy bough]

send it once and wait, it doesnt need to be the latest message

[simple kayak]

hello, i need help. i tried using rvc, but i have a problem,when i press start, in my CMD it always says Pipeline is not initialized, and my sound doesn't change at all, but when i use passthru my sound appears but it's my original sound, i'm using cuda_v.1.5.3.15 version. currently i'm using Amd Rx 570 8gb

[hallow thistle]

You forgot which message you should reply to? The W-Okada version you're using is the original version, which is outdated in the end. AMD Radeon RX 570 will work with W-Okada, but it won't be that fast if you wanna use it with a game.

If you have an AMD GPU, download and use this better W-Okada instead. https://rentry.co/ForkVoiceChangerGuide#download-for-amd-intel-and-cpu-on-windows

[hallow thistle]

Wow, my Kaggle tutorial is lost, so I think I have to teach someone on how to use Kaggle again.

[deft pewter]

GTX 1050 I used a fork version and I didnt like it tbh

And also this never happened before

Been using it for so many months just started getting this issue now

[latent kettle]

I know someone

@royal jacinth

[hazy belfry]

Okay let’s talk then @latent kettle

[latent kettle]

Check your dm

[hazy belfry]

Ok

[simple kayak]

what should i do next after extracting it?

[hallow thistle]

You didn't like it? If you used the older version of fork W-Okada, it could happen. Or else something is wrong with your PC.

[devout slate]

How do I actually make a full on ai.

[deft pewter]

how? i haven't even installed anything

while using w-okada

it just randomly happened

[eternal aurora]

Okay

[deft pewter]

@hallow thistle the cmd is this right before it closes:

using default ngrok token.
checking the modules...
2025-04-27 13:38:15 | INFO | ngrok.session | Session created with auth token
2025-04-27 13:38:15 | INFO | ngrok.listener | Created listener "cd85d7f8982c8c703bb7114ff6d3bdff" with url "https://private-------.ngrok-free.app"


2025-04-27 13:38:15 | INFO | ngrok.listener | Listener "cd85d7f8982c8c703bb7114ff6d3bdff" forwarding to "tcp://localhost:18000"
                        vcclient
2025-04-27 13:38:15 | INFO | ngrok.tunnel_ext | forward_tunnel; tunnel_id="cd85d7f8982c8c703bb7114ff6d3bdff" url=tcp://localhost:18000

----------------------------------------------------------
 Application | http://localhost:18000
 Log(rich)   | http://localhost:18000/?app_mode=LogViewer
 Log(text)   | http://localhost:18000/vcclient.log
 API         | http://localhost:18000/docs
 License(js) | http://localhost:18000/licenses-js.json
 License(py) | http://localhost:18000/licenses-py.json
 Ngrok       | (private-----ngrok)
----------------------------------------------------------
Please press Ctrl+C once to exit vcclient.

and i'd still have the voice changer open but would be broken too

[hallow thistle]

This is the original version of W-Okada. I use fork W-Okada. Original and fork W-Okada are whole different looks..

[deft pewter]

i know. most of the stuff dont load becaue the batch terminates its self for some reason before it finishes loading and even if it finished loading voice changer models wont work nor would u be able to switch

fixed it.

[hallow thistle]

Do not run W-Okada through batch. Run W-Okada on CMD instead or double click on the program.

[modern swallow]

Does W-okada on colab work ok?

[hallow thistle]

Original W-Okada notebook is broken. Using fork W-Okada on your free Colab can get your account terminated.

[modern swallow]

what should i do to use w-okada =v

[hallow thistle]

Running W-Okada directly on your PC and Kaggle are the only hopes.

[modern swallow]

my pc uses intel R UHD Graphics 620 gpu☠️

[hallow thistle]

That leaves Kaggle your only hope now, until you buy a better PC with better GPU.

[modern swallow]

but i don't know how to use kaggle

[modern swallow]

now i know how to use it🤓

but how to use it on discord=v

[deft pewter]

just use virtual cables

input (your microphone) output (virtual cable you have), monitor (your speakers [this is optional its only to monitor your self])

[modern swallow]

why is it not working =v

[deft pewter]

did u give it perms

[zealous shoal]

why does coalab keeps disconnecting me

im using a diff acc

[low shard]

Elaborate

What's your PC GPU? What you want to do? What google colab link are you using?

[zealous shoal]

that message thing that pops up in colab

[knotty moth]

more likely outdated colab that Google is banning it

[low shard]

Yes, elaborate what I asked you

[zealous shoal]

Runtime disconnected
Your runtime has been disconnected due to executing code that is disallowed in our free of charge tier. Colab subsidizes millions of users and prioritizes interactive programming sessions while disallowing some types of usage as outlined in the FAQ. If you believe this message is in error, file an appeal. Please include any relevant context about your usage.
Your compute unit balance is 0. Purchase more

To connect to a new runtime, click the connect button below.

[low shard]

Yes, elaborate what I asked you

[zealous shoal]

t100

[low shard]

What's your PC GPU? What you want to do? What google colab link are you using?

[zealous shoal]

i meant t4

this one

https://colab.research.google.com/github/deiteris/voice-changer/blob/master-custom/Colab_RealtimeVoiceChanger.ipynb#scrollTo=lLWQuUd7WW9U

[knotty moth]

it is colab's, what's yours?

[low shard]

That's google colab GPU, not yours

[low shard]

Always check #📰│dev-updates

That one gets detected by google

Bc you can't use web UIs in Google colab free tier

And also original wokada is broken on colab

So u less you pay you can't use wokada on colab but there are other ways

[zealous shoal]

so is there no ui one?

[low shard]

@zealous shoal first of all, use task manager to tell your PC GPU

Maybe you can do it locally if it's good enough

[zealous shoal]

i have a 780ti

[low shard]

Damn

Cooked

Well local isn't your option then

[zealous shoal]

thats why im going for cloud type

is there any ways? no payment

@low shard

[low shard]

Only Kaggle, 30 hours weekly but it's harder to use and also you need to have a Google account and verify your phone number

https://docs.aihub.gg/w-okada/cloud/w-okada-kaggle/ this is the guide for the wokada deiteris fork kaggle

[zealous shoal]

the acc isnt a problem

[low shard]

Reminder that you need also to verify your phone number else it won't work

[zealous shoal]

alr

[low shard]

If you don't want to, your only other option is paying for colab

Cloud isn't stable

[zealous shoal]

damn

[low shard]

It gets broken easily

It's better having a good PC and do it locally

[zealous shoal]

i tried doing locally it did a bsod

on my 780ti

[low shard]

Because the providers update the dependencies

[low shard]

Of course, it has an ancient GPU

That's why I highlighted having a good pc

The bare absolutely minimum is a GTX 900, but your PC is even way weaker than that

[zealous shoal]

well the 980 performs 11% better than 780ti

[low shard]

Summarizing, your only options are:

• buy a new PC
• use Kaggle
• pay for colab
• wait months if not more for colab to get fixed, if it ever does for wokada

[low shard]

The 780ti doesn't even get recognized by CUDA anymore which is needed for AI

It's ancient

[zealous shoal]

aw man

[low shard]

AI is an intensive task

You can't expect to run it on ancient hardware

[low shard]

@zealous shoal well, let me know

[zealous shoal]

dang i cant verify a number

[low shard]

If it's giving issues, contact Kaggle support and wait till they email you back

[low shard]

Also no, it needs a user interface

[zealous shoal]

alr, ill wait till everything becomes free

wait will a p4000 work?

[low shard]

Good luck, it's like this since 4 months

[zealous shoal]

i think my friend has one

[low shard]

You can't expect everything for free

[zealous shoal]

@low shard

[low shard]

Not sure but worth a try

[zealous shoal]

wait about the t4 gpu

does it require display?

ill save money for it

i have an extra pcie x16

[low shard]

@zealous shoal Hina is busy and can't make original Hina mod Wokada to work, and no matter what we do Wokada deiteris fork get detected, soo it could be like this forever

I would just suggest you to use Kaggle support, they will reply to u TMR prob

[low shard]

Wdym? You don't have a monitor?

[zealous shoal]

yes i do

[low shard]

It doesn't use a regular UI, it uses a Web User interface that opens in your browser

[zealous shoal]

bc it saids on the nvidia website, you can have two gpus

one for ai

and one for any task

[low shard]

I mean you can just use that GPU for ai yeah

Tho you should prob take in consideration that your main GPU is old and your other use cases of this PC other than using this program

[zealous shoal]

me too poor

[knotty moth]

RTX 2060 could be at most affordable for you

[low shard]

Just contact Kaggle support https://www.kaggle.com/contact

[knotty moth]

or better, 3060

[low shard]

You don't even need to write an email to contact them for ur phone number not working

[low shard]

They have a 780ti

[knotty moth]

I mean the upgrade suggestion

[mossy sedge]

What to do if the microphone is not selected in the voice changer

[low shard]

elaborate:

• your pc gpu
• what you want to do
• what tutorial link are you using
• whats your browser

[knotty moth]

#1366037225707405444 message

[low shard]

they are deffo using a yt tut 😭

well better to use that support forum

[mossy sedge]

GTX 1050 TI Laptop

[low shard]

replied already in the post forum

[surreal cobalt]

hi im frands Meow:3

[brittle wing]

yo can anyone help me? i have problems with ookada

[knotty moth]

any problem you need to get help on?

[knotty moth]

you too, pls explain it

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[fading barn]

Is there a guide to install on linux mint? I was using https://rentry.co/ForkVoiceChangerGuide#linux but it says to use "sudo yum" and yum isnt a command on mint

[low shard]

elaborate:

• what's your pc gpu
• what u want to do
• what tut link are u using
• what is the issue

[low shard]

maybe try apt instead of yum?

I don't use linux

[fading barn]

tried that, it just says unable to locate package portaudio

Same, this is my first time really trying it haha

[low shard]

what if u try sudo apt install portaudio19-dev ?

[fading barn]

That seems to have worked, I'll continue the guide and see what happens

[low shard]

let me know, just updated the guide

[fading barn]

Oh well I guess there wasn't much more to continue for installing, just need to figure out how to open it now haha

Like if I double click it, it generates the pretrain folder and some files, but nothing actually pops up

[low shard]

doesn't it show a terminal window?

it's normal that it generates internal files, but it should show that it's doing that in a terminal window, atleast on windows

maybe wait and see if it opens the browser alone when it's done generating files

[fading barn]

Nope, I am kind of getting it though, I opened terminal myself and used chmod +x, then just ran ./MMVCServerSIO now it's doing some stuff

chmod +x may have been unnecessary but oh well

[tough fiber]

is collab doesnt work anymore?

https://colab.research.google.com/github/deiteris/voice-changer/blob/master-custom/Colab_RealtimeVoiceChanger.ipynb#scrollTo=lLWQuUd7WW9U

this one?

/bin/bash: line 1: ./ML_Program: Is a directory
WARNING:pyngrok.process.ngrok:t=2025-04-27T14:45:00+0000 lvl=warn msg="Stopping forwarder" name=http-18888-5b0d435b-886b-4599-abea-a0f13e8760db acceptErr="failed to accept connection: Listener closed"
--------- SERVER STOPPED! ---------

[fading barn]

Cool it works now! I got the error "MMVCServerSIO/_internal/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by /lib/x86_64-linux-gnu/libjack.so.0)
[PYI-26775:ERROR] Failed to execute" after it finished installing everything, so I went to the _internal folder and renamed libstdc++.so.6 to libstdc++.so.6 .bak, then ran "sudo apt update
sudo apt install libstdc++6 libportaudio2 jackd2"

now it works fine

[low shard]

the original hina mod colab is broken

the one you're using, wokada deiteris fork gets detected, it works only if you pay colab because web uis aren't allowed on the free tier

@tough fiber
what's your pc gpu and what do you want to do?

[woeful obsidian]

Does anyone had success with rvc webui on amd gpu using dml

If i use the req-dml.txt i can get dml working but the conversion always gives a error about not finding the source file

Its a wav file i used fine on the cpu version

Windows 10

[low shard]

elaborate:

• whats ur pc gpu
• what u want to do
• what tutorial link are u using

[woeful obsidian]

6700xt
Use ai models to make covers and maybe train my own model

As far as tutorial i asked chatgpt

Downloaded the latest build for amd from github
Made virtual environment
Activate it
Installed torch dml
And dml.txt
Made sure to use python 3.10.9 and pip 23.3

[low shard]

As far as tutorial i asked chatgpt
chatgpt doesn't know about RVC

your pc gpu is good enough

but the original RVC(also known as mainline), doesn't have windows amd support

it only has linux amd support

@woeful obsidian what's your OS?

(Operative System)

[woeful obsidian]

Windows 10
Thats why i used the directml
Which works on windows
Its not as good as nvidia but much faster than cpu

[tough fiber]

3070 but on cs2 its pretty bad

[low shard]

the mainline rvc doesn't have windows amd support, only linux with rocm

it's shown in the readme

you need to use applio (rvc fork) with zluda (cuda emulator) https://docs.applio.org/applio/getting-started/installation#amd-gpu-support-windows

AMD support isn't that well for AI, especially on windows where you gotta use an emulator lol

[low shard]

you don't even need to use cloud (remote good pc), your pc gpu is good enough to run it locally

you'd just need to set the graphics to the lowest for games

check the wokada deiteris fork https://rentry.co/forkvoicechangerguide

[woeful obsidian]

Im confused

for Nvidia graphics cards
pip install -r requirements.txt

for AMD/Intel graphics cards on Windows (DirectML)：
pip install -r requirements-dml.txt

From readme

[low shard]

okay nvm it was in the other readme inside the docs

soo, can you share a screenshot of the error?

we usually suggest applio anyways, because it has better performance and support (generally speaking)

along with an easier User Interface

[gilded flicker]

Anyone know how to run hugging face models?

[low shard]

elaborate

what's your pc gpu? what you want to do exactly? what models are yout alking about?

huggingface is one of the biggest site that contains AI models of all type, RVC, image to video, text to image, etc etc

along with also datasets and other things

[woeful obsidian]

Performance as in quality or time to convert
Does it have all the modes of rvc ?

[low shard]

performance as in speed mostly

wdym with all modes? maybe you meant models? yes ofcourse it can run all rvc models

[woeful obsidian]

No i meant algorithm i guess
There is like 4 or 5 to choose from in rvc forgot the names

[low shard]

wdym algorithm

it's still rvc, it runs rvc models fine

[woeful obsidian]

I mean pitch extraction algorithm
Like pm rmvpe

[low shard]

yes ofcourse it has rmvpe

rmvpe is the best one

the others one are there but aren't suggested at all

[analog obsidian]

model quality depends on the dataset

the fastest f0 is fcpe, it uses 1gb of vram in fp32 mode
but it's a bit unstable

[woeful obsidian]

I know
So this applio and all its dependcies have all been scanned and safe?
Im paranoid about that stuff lol

[low shard]

it's safe yes, it's been a fork since like a year and it's open source

the code is public at https://github.com/IAHispano/Applio

[analog obsidian]

yea dont use mainline if you got an amd gpu

[woeful obsidian]

Does it do anythinf better than main rvc or just more user friendly

[low shard]

better performance, better amd support via zluda and more user friendly

it's the most suggested one for amd

[woeful obsidian]

Ok dope
Thanks

[spare solstice]

i want to contact mods

can someone help?

[gilded flicker]

Can I dm you about it

[spare solstice]

@half olive

hey sorry to ping there's an issue

[low shard]

No need to, you can elaborate here or in #1192011222023950368

I already asked you the things you need to elaborate

[spare solstice]

nick are you a mod

[low shard]

Hello, are you looking for tech help or something related to moderation?

[low shard]

I am a Junior admin, and mod yes

[spare solstice]

i guess moderation so there is a user who is scamming people from this server

i have screenshots if you need

[low shard]

please dm @vital hedge so it will let you open a ticket on the server and the situation can be seen by the whole staff team

[spare solstice]

what you mean i have screenshots why is there a process to whole thing lol

i thought i should let you guys know i am not dumb enough to fall for it but maybe someone will you know

[low shard]

sorry but we do this so we can check the situation as a team and have the ticket saved for moderation proof, it takes only 2 minutes don't worry, you don't need to do anything else than just dming the bot and showing the info in the ticket

[spare solstice]

oh thank you so much nick i did it ticket is generated i hope it resolves soon, thank you again bro ❤

[carmine fox]

Where do I find the latest and correct version for the voice changer?
When I launch mine now, I goes up to like 30k ms...
I have AMD and Windows 11

[latent kettle]

What happened ?

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[latent kettle]

@carmine fox

Latest as of 7th December 2024

[carmine fox]

Thanks a lot!

[low shard]

what's your pc gpu btw

[carmine fox]

Any idea what a good female voice is for someone with an accent? xD

[low shard]

i'm guessing realtime for calls

[carmine fox]

7900 XTX

Yep!

[low shard]

great

[low shard]

yeah then just use the 1st link, it's the wokada deiteris fork which is the most optimized and has alot of improvements generally, but even for amd

[carmine fox]

Yep, I just need to test with a good model now I suppose

[low shard]

be sure to download vac lite at the 3rd step and the wokada deiteris fork amd windows version

[carmine fox]

My last fork began spiking up to 30k ms, I dont know why

[low shard]

if you had vb audio cable, uninstall it as it causes issues reported by other users

[carmine fox]

I think its vac lite I have right now

Any way to check?

[low shard]

alr, do u may want me to check ur settings?

[carmine fox]

I would love that

[low shard]

it will be named as "Line 1" in the wokada output options

!give-media-perms 1h @carmine fox

u can share screenshots now

[carmine fox]

Yep!

This is lite?

[low shard]

yep

u can share a screenshot of the rest of ur wokada settings too

[carmine fox]

Well it's default right now, one sec:

[low shard]

gpu: your amd gpu

extra: 2.7
chunk: 128

set input as your microphone

also, you should be able to use force fp32 mode in advanced settings in amd too, it makes the models more stable and better quality at the cost of some delay, your choice if you want to use this

[carmine fox]

Yes, perfect!

[low shard]

you could optionally also use https://rentry.co/forkvoicechangerguide#reduce-more-delay for less delay, but you won't be able to use echo & noise suppression and it's a bit more complex

[carmine fox]

The issue I always had with Okada is how you can just naturally tell the voice isnt real, bugs out too at certain tones. But it could just be my settings or setup? I never heard from someone who is actually good at this

Is there a good, realistic female voice you can recommend? For someone who is going to speak English, but has a slight accent

[low shard]

The settings you're using right now are good, I would suggest you to try models with those settings

if you still have the same issues, it might be the model aren't properly trained, like at a specific pitch range, since those are models trained by the community, there aren't like 'official rvc models' since it's an Open Source project
it would be better you play around with the pitch option and try other models

[low shard]

https://rentry.co/forkvoicechangerguide#voice-models-to-try-out might help you out

or if you want to search them yourself, You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @earnest musk
• https://weights.com/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.aihub.gg/essentials/how-to-make-voice-models/

[earnest musk]

:wave: @low shard, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[carmine fox]

Thank you so much, I really appreciate the help

[low shard]

you're welcome, need any other help? or is this all solved?

[carmine fox]

Im actually getting the same issue with this FORK as the last...

ms spiking up to infinity

And it just keeps spiking

It stopped now at 36k ms

[low shard]

36k? The perf value showed only 288

[carmine fox]

Look at the ping

[low shard]

Are you using it with a game too?

[carmine fox]

Not now nope

I was planning to tho

[low shard]

Can you close every useless program in background, then put the chunk to 400?

[low shard]

Just remember to set the graphics to the lowest, and if the perf value goes red, put the chunk a bit higher than it

[carmine fox]

I have nothing open now except my browser and discord

Still the same issue

I never had this issue before

Guess I can try restart my PC

[low shard]

Like, you keep rising the chunk, and then the perf goes even higher and red?

[carmine fox]

Seems to be fixed at 400

[low shard]

Did restarting fix this? Or just increasing the chunk?

[carmine fox]

Just increased the chunk

[low shard]

Yeah, I think the issue was related to the perf being red

Be sure that the perf is always green, else increase the chunk

Any other issues?

[carmine fox]

Should be good thanks! Last question, should I keep all the volumes at default? 100%?

[low shard]

Should be fine, increase them only if you hear yourself too low

Also, monitor is optional, it's only to hear yourself if you set it to your headphones

[carmine fox]

How do I enable the monitor audio?

Do I just put it to my headset?

Yeah that fixed it

[low shard]

Yup

[carmine fox]

Thank you so much! I think I am good from here

I really appreciate your help

[low shard]

You're welcome, have a nice day

[eternal aurora]

for the next step after setting up applio and tensorboard

what do i do from here?

nvm

where do i do this?

[low shard]

that's for applio colab not applio kaggle btw

[eternal aurora]

yeahh ik idk how to do this in kaggle

oh wait

nvm

i didnt even realise i pressed the collab one omd

[eternal aurora]

the only issue is idk what to do from here

[sharp lava]

im trying to downlaod cuda and all that for voice training im on the step where i downlaoded cudnn and powershell is aking my to find the path, but im like so lost and cant find any of the files i donwlaoded it, would anyone be able to help me

[eternal aurora]

im doing batch right? since i got multiple files?

[low shard]

lemme guess, you're following a youtube video tutorial?

[low shard]

wdym

[eternal aurora]

nvm i found the guide

[eternal aurora]

basically im here right

[sharp lava]

nope

[eternal aurora]

and i have got multiple files however its not letting me put the files in

[low shard]

elaborate:

• your pc gpu
• what you want to do
• what tutorial link are you following

[low shard]

oh yeah batch inference is just inferencing multiple files, you just need to put the folder that contails all files

be aware that it's not much suggested since you might need to play with the pitch on some specific files

[vernal belfry]

hey! Im using APPLIO COLLAB and my dataset is 25 min, and when i try to pretrain it says the track lenght is 00:00
do i need to cut my wav. into 2 parts or what should i do?

[grand vapor]

Hi everyone

Im searching ai voices library

[low shard]

elaborate:

• your pc gpu
• what you want to do
• what link are you using
• a screenshot of the error

[eternal aurora]

oh-

[low shard]

do you need any help

[eternal aurora]

when you say i need to put it in the folder what folder do i put it in?

i have the files on my pc

and i copied the path for it but its not appearing in the voice model

[low shard]

the input folder is this, go there in the file url

[sharp lava]

i figuired it out and its now dowlnading python but when it started doing thing it moved my cudnn file away from my file i want my rvc in, is that normal, do i need to move it back?

[eternal aurora]

like this?

[low shard]

pls elaborate #✨│ai-help message

[eternal aurora]

or does it need to be in kaggle im guessing?

[low shard]

no, it's on cloud, it's connected to its own pc files not your pc at all

[eternal aurora]

ah shit

[low shard]

you can either upload the files via the UI or use the file url to put the files in that folder

[eternal aurora]

how do i uppload this to kaggle?

[eternal aurora]

how do i use the file url

this>?

[low shard]

it should be given in the kaggle output near to the applio url

[low shard]

yes

[eternal aurora]

yeah i just found it lmao

[sharp lava]

Nvidia geforce rtx 3080 ti
I want to make a voice trainer, using RVC-Project
/
Retrieval-based-Voice-Conversion-WebUI

and ive been follwing just asking questions in chat gpt, and i downloaded cudnn off of nvidia and by using powershell and followed some prompts, and its currently downlaoding python now

[eternal aurora]

btw whats the audio format?

mine is wav

[low shard]

audio extension which can be .wav, .mp3, etc

[eternal aurora]

its not showing up for some reason

[low shard]

Nvidia geforce rtx 3080 ti
great

I want to make a voice trainer,
I guess training a model

chat gpt
chatgpt doesn't know about RVC
You have been trying to use the original RVC (aka mainline), which isn't much suggested to Applio (an RVC fork, modified version)

it's better you uninstall everything you did right now, and follow our docs guide for applio https://docs.aihub.gg/rvc/local/applio/

[vernal belfry]

windows
trying to create and train a new model in Applio Collab
the error says:
Preprocess completed in 0.01 seconds on 00:00:00 seconds of audio.

An error occurred extracting the index: need at least one array to concatenate
If you are running this code in a virtual environment, make sure you have enough GPU available to generate the Index file

No wav file found.
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1745779166.624992 3045 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1745779166.631806 3045 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
/usr/local/lib/python3.11/dist-packages/torch/utils/data/dataloader.py:558: UserWarning: This DataLoader will create 4 worker processes in total. Our suggested max number of worker in current system is 2, which is smaller than what this DataLoader is going to create. Please be aware that excessive worker creation might get DataLoader running slow or even freeze, lower the worker number to avoid potential slowness/freeze if necessary.
warnings.warn(_create_warning_msg(
Not enough data present in the training set. Perhaps you forgot to slice the audio files in preprocess?
An error occurred extracting the index: need at least one array to concatenate

[low shard]

windows
That's your OS, Operative System, not your pc gpu

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

Colab is cloud (remote good pc), it's meant only for people with a bad pc

[sharp lava]

so this will allow me to train a voice by me speaking my normal voice and evntually no matter what i say it will sound like what im trying to mimic?

[low shard]

what? you can't train an ai model to work perfectly specifically for your voice

the ai training will work based on the dataset you give it, how good is the dataset, and how you monitor the training with the tensorboard

Also, RVC (and Applio) is for training and inference (use models) on pre-recorded audios, you should check wokada deiteris fork if you want realtime inference for calls or games (along with Applio to train the voice)

[sharp lava]

i think that what i ment, using a model and then traing it for me specifically

[low shard]

no, you can't train a model for it to be perfect only for your voice

you can train a model based on the voice, not to work specifically with a voice

if you just found a model randomly online and try to 'train it' to work with your voice, that's impossible

just find another model, not all models are super high quality, or try to play with the pitch

@sharp lava are u trying to train a model specifically to work with your voice because the realtime voice changer isn't working well? may u want me to check your settings too?

[sharp lava]

then what is RVC-Project, is there no difference

Appilio and RVC-Project what is the differnece

[low shard]

Applio is a fork (modified version) which has better performance and easier to use

there is no single way in any program to train a model specifically to work with your voice

but, you can train a model based on a voice (like peter griffin for example)

[sharp lava]

yeah thats what i mean i dont mean my voice, i just mean like when i talk and train a model itll sound like the model, peter griffern for example would be trained from peter griffen clips and all that

i dont mean my voice specically

[low shard]

you don't just talk into the program and magically it trains itself

[sharp lava]

yeah i know that

[low shard]

you need to make a dataset based on the voice, clean it to be perfect, make sure it's only vocals, check the sample rate, set the correct training settings, then start training and monitorage it with the tensorboard

so yeah, if you're looking for this, this is how RVC training works, and it's the best you use Applio for best performance

[sharp lava]

got it

can you also use Appilio as the voice changer it self, or is it just for training, I currently use W-okada

[low shard]

no, not as realtime voice changer

it's just for training and pre-recorded audios inference

also, are you using wokada deiteris fork?

share a screenshot of ur wokada

!give-media-perms 1h @sharp lava

[brittle wing]

What i need to put it in the F0 Det

[low shard]

elaborate:

• your pc gpu
• what you want to do
• what tutorial link are you using
• whats the error

[brittle wing]

NVIDIA RTX 4090
VOICE CHANGER

[low shard]

you didn't elaborate everything

reply to each thing I asked

@brittle wing do you remember what tutorial link did you use? if not, send a screenshot of the program

[eternal aurora]

Quick question when I want to use Applio do I need to do run all of the cells

[low shard]

!give-media-perms 1h @brittle wing

[low shard]

for kaggle? no just the start ui (last) one

[eternal aurora]

Ahh okay

Do you think 13-14 minutes is enough for the audio?

[low shard]

for training? as long as it's good quality

[eternal aurora]

Yeah I believe so

And would u know anything abt extracting game audio from a game file?

[paper bloom]

hey, do u know why i can hear ppl in the game thru my owakada ?

like its echoing in the game sometimes

and than ppl can hear themselves thru my mic

[low shard]

share a screenshot of ur wokada

[low shard]

u should check the game modding community

[paper bloom]

okay

plus ppl say im to quiet

is it my mic or my settings?

[eternal aurora]

Okay

[low shard]

set output to line 1

set in sens a bit more to the right

if it still persists, you will need to either:

• lower your headset volume and put ur mic away
• use client and use echo sup2

[paper bloom]

output line 1 means i cant hear the changed voice right?

[low shard]

yes, because for that you need to set monitor to ur headphones

[paper bloom]

could u pls explain what monitor means?^^

ohh perfect it worked now after i changed my monitor to my ehadphones ❤️

oh and where do i finde the turorial for the settings?

[cedar mirage]

Hello, I am looking for an RVC file in Bulgarian.

[low shard]

it's basically for monitorage of the voice conversion

so you can both use it to the program and both hear it urself at the same time

[low shard]

https://rentry.co/forkvoicechangerguide#settings

[low shard]

you were the guy who had an rtx 4070

well wdym rvc file in bulgarian? maybe an rvc model in bulgarian?

[paper bloom]

oh but like what diffrent does it make if i put monitoring to my headphones or lite

[low shard]

shouldn't have a difference

it's just meant as a 2nd output to hear yourself at the same time while using the model

[paper bloom]

oh okay

ty

[low shard]

bc if u put the output to line 1, u can't hear urself

yw, so is it fixed?

[paper bloom]

but i cans till ehar myself

i mean atleast the changed voice

but the echoing is fixed i think its a bug sometimes

[low shard]

i think that's just because of echo in your headphones, you're not supposed to hear urself if u had output to line 1 and monitor to none

[paper bloom]

my settings before was output my headphones and monitorm was line 1 i learned that from a youtube turoial back than

[low shard]

you should forget about what you see on youtube for wokada

there's people who suggest even crepe_tiny or harvest on youtube for wokada 😭

[paper bloom]

is this good now?

[low shard]

yes

[paper bloom]

btw where do i found the turoial in the server or do i always have to look into this chat?

[low shard]

#1159513888199540817

just right above the support channels

or https://docs.aihub.gg

[paper bloom]

ty very much

[low shard]

yw

[mystic bobcat]

what app shoudl i use for the voice changer

[proud hinge]

Im having troubles with the Real time voice changer, it keeps buffering and not play the full audio just parts of it how can I fix this. I have a AMD RX 6600 and I want to use the audio in discord or any other games.

[cedar mirage]

yes I am the same I am looking for a file in Bulgarian

[simple ore]

rvc models are not language specific

[eternal aurora]

so i just uploaded the audio to the online file link how do i get the link and put it into the applio input place

[eternal aurora]

how do i paste the linkkk

[simple ore]

what colab, what link, what are you trying to do?

[mighty pewter]

@low shard I have a Geforce 1650 Super right now

Will it be able to run?

[eternal aurora]

Applio on kaggle

So I got Applio all set up

And I need to fill in the input box with a link

I got the Audio on a folder here

But idk how to copy it and paste it into the input

[mighty pewter]

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

It will be fine except for intensive games

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

1st link

[mighty pewter]

I'm so confused on what too download

sorry im new too all this lol

[low shard]

Download vac lite at the 3rd step, then the wokada deiteris fork Nvidia windows version

[mighty pewter]

is there any video too installing this?

[simple ore]

what are you trying to do? run inference? train a model?

batch inference is only if you're able to upload multiple files / local install

easiest way to upload files using UI to colab/kaggle is to go to the training tab and make a dataset

you'll get /something/something/assets/foldername -- that's what you can use in batch inference

[low shard]

No, video tutorials are outdated

[elfin nebula]

I'm trying to install Zluda and I got a fatal error on my cmd prompt
More specifically "fatal: too many arguments"

[simple ore]

show a screenshot

of what you're trying to do and the error

[elfin nebula]

I'm trying to do the setup steps on the github page

[simple ore]

you have extra 'git clone'

if you did not notice

[elfin nebula]

That's directly copied from github

[simple ore]

i'll let the maintainer know

[elfin nebula]

Thank you

Should I install torch? It didn't say anything about doing that...

[simple ore]

no, that's message from attempted uninstall

[elder coral]

How do i use kaggle for training models

[karmic bronze]

how to find the best spanish voices?

and remove the "robotic" sound

[elfin nebula]

Can we use a longer reference audio for a more accurate voice cloning in Zluda

[elder coral]

How to colab

Pls show me links to make voice models in colab

[viscid moss]

-colab

[patent trellis]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[elfin nebula]

Can Zluda use the amd gpu instead of the cpu?

[knotty moth]

do you know what zluda is supposed to be for?

[elfin nebula]

Mb I mean fish speech using zluda

Ngl someone just recommended it to me, and I've been taking shot after shot in the dark to try and make it work

I have 0 idea on what I'm doing

[simple ore]

it does

[knotty moth]

if you haven't, you can search the documentation for that

[elfin nebula]

I tried making a voice clone and have it said 3 words and it took almost 4 minutes, is it possible for it to be faster?

[simple ore]

it is compiling the kernels, gonna take 20 min 1st time

[elfin nebula]

Wait so before I proceed with this, I can use this for Realtime tts right?

Kinda like an ai assistant?

[elfin nebula]

Are you talking about something in the #1159513888199540817 ?

[elfin nebula]

Actually this is the whole thing... Idrk if there's an issue but I think it's worth asking at least

[simple ore]

not sure why it says cuda not available

you probably did not start it properly

[elfin nebula]

So apart from "Cuda not available" everything is fine?

[simple ore]

yeah

[elfin nebula]

So I did relaunch it, I just clicked on the fsz batch file in the folder

Is there another way of opening it?

[simple ore]

no

if that fails it means you not do the preliminary stuff - hip sdk, adding it to the path

[elfin nebula]

I did do that tho

Probably did it wrong then...

I think I did it correctly

Do I have to do step 6 even if my GPU is the 7900xtx? Could that be the issue?

[simple ore]

that looks fine

[simple ore]

no, 7900xtx is supported by default without hacks

[elfin nebula]

Then idk the issue then

[simple ore]

which python version do you have installed?

run cmd.exe, from it run where python

[simple ore]

okay, that's fine too

[elder coral]

I can't import models into huggingface

because the file is too big

[simple ore]

rvc models are like 55mb

hugging face allows uploads up to like 10GB

[elder coral]

yes mine is 49mb but it can't be importyed

[elfin nebula]

This is what my Microsoft Visual installer looks like when I launch it, I tried repairing and reset my pc still the same issue

[elder coral]

maybe is it depending on my internet connection

Or it contains malware

[simple ore]

let me just install it myself

[simple ore]

it seems there's a missing line in install-amd.bat

python -m pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu118

[knotty moth]

that is the pth file alone, with index file included in the zip it can be more than 500 MB

[simple ore]

@elfin nebula it should be after pip uninstall torch torchvision torchaudio -y --quiet

[knotty moth]

you'd have to reupload each of the pth and index file in the space

[elfin nebula]

Wait so what should I do?

Do I run that in cmd?

[simple ore]

no, edit the install-amd.bat

add the line I mentione after that line

delete venv

then re-run install-amd.bat

[elfin nebula]

Okay this is where I get confused. Where do I edit that?

[elder coral]

What does An error occurred connecting to Discord: Could not find Discord installed and running on this machine. mean

[knotty moth]

the cloud environment doesn't have discord within it

mostly only in local system with discord being run

[simple ore]

open .bat file in notepad

insert the line, save

[elfin nebula]

Ah gotcha.

[elder coral]

Malware

[elfin nebula]

I can't find the .bat file

[simple ore]

you can always just nuke the folder and start over

[elfin nebula]

Nvm I found

So do I jsut run it like normal now?

[simple ore]

yes

[elder coral]

Bro how do i access the gradio with ngrok

[elfin nebula]

Seemed to work but it's using the wrong gpu

Or is it saying that it's disabling that gpu

[simple ore]

need to set HIP_VISIBLE_DEVICES env variable to either 0 or 1

then restart

[elfin nebula]

Okay where do I insert that?

[simple ore]

in either system or user environment variables

or in fsz.bat

set HIP_VISIBLE_DEVICES="0"

[elfin nebula]

Seemed to work in fsz

[elfin nebula]

How do I "train" the voice model for it to mimic as accurate as possible?

[smoky linden]

Dose anyone know how to install rvc i forgot

[eternal aurora]

Okay

[brittle wing]

whats the link for virtual audio cable

not vb cable

[hallow thistle]

https://software.muzychenko.net/freeware/vac470lite.zip

[hallow thistle]

Which RVC? Applio or W-Okada?

[elder coral]

Does this model sound

[eternal aurora]

Run interface first

But I understand wym

Would it be good
If I combine all of my
Audio files?

[woeful obsidian]

in applio where do i put the pth and index files

[soft zealot]

can someone help me

why does the llm break starting from the second input message

it starts spouting nonsense

import os
import logging
from fastapi import FastAPI
from pydantic import BaseModel
from transformers import AutoProcessor, LlavaForConditionalGeneration, TextIteratorStreamer, BitsAndBytesConfig
import torch
import asyncio
from collections import deque

# FastAPI app setup
app = FastAPI()

# Define the incoming request schema
class SpeechInput(BaseModel):
    speaker: str
    text: str

# Setup logging
logging.basicConfig(level=logging.DEBUG, format="%(asctime)s [%(levelname)s] %(message)s")

# Constants
MODEL_ID = os.getenv("LLAVA_MODEL_ID", "CreitinGameplays/pixtral-1")
QUANT_CONFIG = BitsAndBytesConfig(load_in_8bit=True)
HISTORY_LEN = 20

# Initialize model and processor
logging.debug("Loading model and processor...")
try:
    model = LlavaForConditionalGeneration.from_pretrained(
        MODEL_ID,
        quantization_config=QUANT_CONFIG,
        device_map="auto"
    )
    processor = AutoProcessor.from_pretrained(MODEL_ID)
    processor.patch_size = model.config.vision_config.patch_size
    processor.vision_feature_select_strategy = model.config.vision_feature_select_strategy
    logging.debug("Model and processor loaded successfully.")
except Exception as e:
    logging.exception("Failed to load model and processor.")
    raise e

# History buffer for chat
chat_history = deque(maxlen=HISTORY_LEN)

# Function to generate a response asynchronously
async def generate_response_from_llm(text: str):
    """
    Generates a response from the LLM using the provided text input.
    """
    logging.debug(f"Generating response for: {text}")

    # Define the system prompt
    system_prompt = "You are Pixel, a cute little AI girl. You livestream sometimes and change moods occasionally. Your responses are rather short and never longer than two sentences."

    # Prepare input for the model
    conversation = [{"role": "system", "content": system_prompt}]
    conversation.append({"role": "user", "content": text})
    conversation.extend(chat_history)  # Add history if necessary

    prompt = processor.apply_chat_template(conversation, tokenize=False)
    inputs = processor(text=prompt, return_tensors="pt").to(model.device)

    for k, v in inputs.items():
        if isinstance(v, torch.Tensor) and v.is_floating_point():
            inputs[k] = v.half()

    # Generate a response from the model using async IO to offload blocking tasks
    streamer = TextIteratorStreamer(processor.tokenizer, skip_prompt=True, skip_special_tokens=True)
    generation_kwargs = {
        **inputs,
        "streamer": streamer,
        "max_new_tokens": 512,
        "do_sample": True,
        "temperature": 0.7,
        "repetition_penalty": 1.1,
        "top_p": 0.95,
        "top_k": 50
    }

    try:
        # Perform inference asynchronously
        response = await asyncio.to_thread(model.generate, **generation_kwargs)

        # Process the response from the streamer
        response_text = "".join([new_text for new_text in streamer])
        logging.debug(f"LLM response: {response_text}")

        # Append model response to history
        chat_history.append({"role": "assistant", "content": response_text})

        return response_text

    except Exception as e:
        logging.error(f"Error generating response: {e}")
        return "Sorry, I couldn't generate a response."

@app.post("/process_speech/")
async def process_speech(input: SpeechInput):
    """
    Endpoint to receive speech input, generate a response from the LLM, and return the response.
    """
    logging.info(f"Received speech input: {input.speaker} - {input.text}")

    try:
        # Generate a response from the LLM
        response_text = await generate_response_from_llm(input.text)
        logging.info(f"Generated response: {response_text}")
    except Exception as e:
        logging.error(f"Error generating response: {e}")
        response_text = "Sorry, there was an error generating a response."

    return {"status": "success", "response": response_text}```

[safe echo]

why is this always flagged?

[knotty moth]

I recall another helper staff showed workaround for apple quarantine thing, try searching it

[safe echo]

but its infected file

[knotty moth]

make sure you don't have some other viruses in your system and try redownloading it

then do the workaround on apple quarantine

[safe echo]

yea but my pc never had this issue or pop up but did as soon as i downloaded so i dont have a virus otherwise.

[low shard]

false positive, the app source code is open and public to see at https://github.com/deiteris/voice-changer

[low shard]

that’s not apple quarantine, macos doesn’t run .exe

[knotty moth]

that sus antivirus tryna be like apple

[low shard]

How?

https://rentry.co/forkvoicechangerguide#mac apple quarantine would say that pytorch is damaged

[knotty moth]

or avast?

[simple ore]

you have a shitty antivirus then

[odd isle]

how often does the voice changer get updated?

[lyric geode]

@knotty moth hai I'm trying to make a Neco Arc song cover but huggingface only let me make one song & it wasn't the voice model I wanted. Do u know how I can make more covers?

This site looks so complicated I don't know what it's talking about when I go to pricing

[lyric geode]

Ok I duplicated to generate privately but now it's saying I need to upgrade Gradio SDK & IDK where the options at

[upbeat carbon]

heya! i was just wondering if you can make a voice model just from a laugh or is it not possible?

[tough fiber]

anyone know this problem

[viscid moss]

not possible

[viscid moss]

What program are u trying to run? And what's ur PC GPU? NVIDIA 50 serie ig

[knotty moth]

seems more like possibly unsupported old gpu

[simple ore]

1. unsupported gpu or 2) trying to infer 1hr+ long file

[knotty pollen]

Selam

[charred vessel]

I got issus on google colab its not downloading the files it says it does not work on the py Version while downloading the Modules after that i tierd using ngrok it didnt even install the pyngronk Module cuz it does not work ?

[hallow thistle]

RVC voice model can be trained on any audio. But with just laughing track, the voice model result would be just laughing sounds.

[upbeat carbon]

thanks for telling!😊

[eternal aurora]

how do i stop iy?

nvm

@low shard i managed to increase my audio from 12mins to 38 mins is that fine there may be repeats but idk

[low shard]

repeats wouldn't really help the training

[knotty moth]

repeating the dataset wont help

[low shard]

it's better you get as much high quality, non repetitive audio as you canr

[upbeat carbon]

heya! how can i make this laugh into a voice model? idk i'm just bored https://voicemod.link/item-asylu-e903-tn

ohhhh okay

what if i want it to be low quality for fun?

[low shard]

that was actually a reply for the other user lol

[upbeat carbon]

oh sorry

mb silly me

[low shard]

well, you can just get as much shitty data you want then follow https://docs.aihub.gg to train that laugh model

trust me it wouldn't be the craziest, we had models made out of toothbrushes and planet sounds 😭

[upbeat carbon]

thank you and i'm sorry for that message i didn't know🙏

[low shard]

it's fine dw, you're welcome

[upbeat carbon]

hey uh do i need ultimate vocal remover for this?

[low shard]

that's for spearating vocals and instrumentals, i don't think that laughing is in a song, so i don't think so

btw don't expect it to sound good, it will mostly be just a 'shitpost model'

[upbeat carbon]

okie i think i got this now...

and also i wanted to do this since someone made the minecraft villager just from sounds and it kinda sounded like he was speaking english

but ik that the laugh is just going to be goofy but i still wanna try!

for fun

cuz i'm bored

[knotty moth]

tbh the minecraft villager model still makes more sense than some sans or spamton model

[upbeat carbon]

fr

i still wonder how it sounds english even tho it's just sounds and the punching noise..

[knotty moth]

you can even try finding some cat model

[upbeat carbon]

cat model?

[knotty moth]

minecraft one perhaps

[upbeat carbon]

okie!

[eternal aurora]

Oh

[eternal aurora]

Oh okay so that means I prolly will get 28 mins

[tough fiber]

yea rtx 50 series

didnt nvidia updated cuda drivers for 50 series?

months passed and rvc cant work on 50 series still thats suprised

[knotty moth]

RTX 50 series need cuda 12.8 which doesnt work on torch 2.6/older

[viscid moss]

Applio? if so u should follow these steps
#✨│ai-help message

[tough fiber]

w-okada

voice changer

optimized version

[tough fiber]

so on voice changer theres no way to work?

[viscid moss]

oh, in that case

https://rentry.co/ForkVoiceChangerGuide#download-for-nvidia-rtx-5000-series-gpu-on-windows

[knotty moth]

there is a version exclusively for 50 series

[tough fiber]

oh thanks

really

[viscid moss]

Can I send you a DM?

i was looking for someone with an 50 series GPU

[low shard]

recruit them 🔥

[viscid moss]

UVR5 UI update will be tomorrow, finally i got the korean translation but I'll add more stuff

[low shard]

Nicee

[viscid moss]

Like... 50-series GPU detection and installation of the required pytorch version, and perhaps a more robust way to update UVR5 UI

[low shard]

yeah that's nice

will help the few people who actually bought the 50 serie for using it rather than scalping

[simple ore]

you can just go ahead and use torch2.7.0 cu128

it's been officially released after all

[tough fiber]

https://media.discordapp.net/attachments/1366414893657620571/1366470341798658171/image.png?ex=6811102d&is=680fbead&hm=d609e277a021f4a47df4d800aacdab450a438fd38e29e754d218245a9caabb06&=&width=1201&height=440 for rtx 50 series gpu this error

[sage plover]

hey guys, is it possible to run gpu intensive games with voice changer? if so how? i heard that you can use dual GPU but is there any other way?

[low shard]

what's ur pc gpu

[low shard]

did u use the 50 serie version

[sage plover]

nvidia 2060 super rtx

[tough fiber]

yea this things happens on 50 series version

i think its cuz of server use instead client thats why

[split knot]

got the same problem too

[tough fiber]

switch client instead server

[split knot]

alright i have 50 series gpu and this error is appearing
object has no attribute 'vc_pipeline'

[split knot]

yeah I did thanks but this error above is spamming the cmd prompt

[ancient ocean]

File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/dataloader.py", line 631, in next
data = self._next_data()
^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/dataloader.py", line 1326, in _next_data
return self._process_data(data)
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/dataloader.py", line 1372, in _process_data
data.reraise()
File "/usr/local/lib/python3.11/dist-packages/torch/_utils.py", line 705, in reraise
raise exception
FileNotFoundError: Caught FileNotFoundError in DataLoader worker process 1.
Original Traceback (most recent call last):
File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/_utils/worker.py", line 308, in _worker_loop
data = fetcher.fetch(index) # type: ignore[possibly-undefined]
^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/_utils/fetch.py", line 51, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/torch/utils/data/_utils/fetch.py", line 51, in <listcomp>
data = [self.dataset[idx] for idx in possibly_batched_index]
~~~~~~~~~~~~^^^^^

An error occurred extracting the index: [Errno 2] No such file or directory: '/content/Applio/logs/JONY/extracted'
If you are running this code in a virtual environment, make sure you have enough GPU available to generate the Index file.

[simple ore]

forgot to run extract features step?

[ancient ocean]

It is resume training

[simple ore]

restore backup?

[ancient ocean]

Everything was fine until 13 april (last backup time). Today i see this. If I not mistaken such folder not been

[ancient ocean]

Yes I repeated the same steps

[simple ore]

go to here

and see what it restored

you need to have the google drive mounted when you do the backup restore

you can also check your files in the google drive

[ancient ocean]

Drive folder

Applio folder (target)

Drive page

[simple ore]

ah.. now you're using 3.2.9..

rename the v2_extracted folder to extracted

[ancient ocean]

really?) what was wrong with "v2_extracted"?

[simple ore]

since we dont do v1 training in Applio, the folder has been renamed

also may need to edit filelist.txt as well

actually it is strange, it should be using the filelist with whatever the filename is there

[ancient ocean]

Yes now it shows error related to missing files in v2_extracted ))). I need to download filelist and use VS Code to replace everything

[simple ore]

ah

I see

it was the mute files

make sure both mute and your model have the folder named extracted

and then change all v2_extracted in the filelist to extracted

[ancient ocean]

mute not included to backup. It created automatically. I just need to change backup and reset runtime

[acoustic thunder]

Is there a difference between Okada and RVC Webui?

[low shard]

in terms of realtime yeah

tell your pc gpu and what you want to do

[young halo]

Can someone pls send a link to a covers inference?

[acoustic thunder]

what is the difference between w-okada and deiteris' optimized W-Okada RealTime Voice Changer Client (Fork)

i have rtx 4060 laptop, just want to do high quality RVC

Oh, i see, overall just better rounded fork. Will try it out

[eternal aurora]

@low shard ngl i have no idea on what im doing

its not even letting me convert

@simple ore

[simple ore]

why did you not select a voice model at the top?

[eternal aurora]

It wasn’t giving me an option to do so I refreshed asw

Unless there’s something else I need to do

[wraith cipher]

is it possible to make the voices sound more excited when I make them?

Just using weights website

[elder coral]

Yes my model has been rejected i need to put all the files

If i train longer my model might be overtrained

[elder coral]

I think this contains malware

[craggy bough]

wrong

nxdomain means the website doesnt exist

[wild canopy]

hi

[elder coral]

I understand ngrok but when i load up the site it says this site can't be reacged

[hallow thistle]

If the "website is malicious", it would throw up a red warning instead of a sad paper.

[craggy bough]

correct, the site cant be reached because it doesnt exist

imagine if you tried to watch a video on youtube and youtube doesnt exist

[knotty moth]

classic 404 moment

[knotty moth]

you might have entered incorrect, nonexistent url, or the ngrok session run through the colab/kaggle cell might have aborted with some error messages

[craggy bough]

cant be reached = malware

[knotty moth]

and any kind of big corps telemetry/backdoors are not malware /s

[elder coral]

Pls bring me back my model maker after you've approved a model that meets the requirements

[knotty moth]

model makers could have their role revoked if fail to meet the QC standard

[elder coral]

why was submit-model replaced with model-maker-role

Where is the QC standard?

So if i train a longer dataset my model will still sound robotic and muddy

[knotty moth]

more likely overly denoising problem

[elder coral]

Where do i read the QC standard

[primal quarry]

whats for the voice models

[crude flame]

https://docs.aihub.gg/extra/model-maker-role/

[unique heath]

Anyone got a good recommendation for a decent voice changer that doesn’t stutter? I tried one, I forgot the name of it.

Specs of my pc is

RTX 4070 TI
And an I7

[karmic bronze]

MMVC i think is the one

[unique heath]

Have you used it?

[low shard]

Realtime right?

For calls/games?

If so

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

1st link Wokada deiteris fork

[hybrid ocean]

How can I use these model voices?

[low shard]

What's your PC GPU and do you want to use them realtime for calls?

[acoustic thunder]

@low shard is there a way to solve this in deiteris fork "RuntimeError: Failed to find NodeArg with name: skip_head in the def list"

[low shard]

Please elaborate a bit more, what did you do and what's your PC GPU

[acoustic thunder]

i press "start" and this error happen. I have rtx 4060 laptop, trying to run RVC model

okay, i reuploaded the model and it's gone

[unique heath]

@low shard so I think I’m currently using Wokada, I’ll have to show you the interface. But sometimes it’ll glitch out and sound very robotic. Or it’ll basically stutter and repeat itself. It’s another real Time voice changer.

[low shard]

are you using the wokada deiteris fork from the guide i sent you?

if u saw a video tutorials, those use an old version of original wokada with vb audio cable, and it’s not good

[unique heath]

Ahhh yes. I use VB audio cable. I’ll check your guide out in a moment! Thank you

I’ll definitely have to try that one out because the one that I currently use is not good whatsoever. I mean, it works fine sometimes but then other times it just glitches out or stutters

[low shard]

vb audio cable gives random issues, and the old original wokada has worse performance

[unique heath]

Yeah, I’ll definitely check out the one that you sent me then. I just don’t like having replay issues,. Thank you, Nick.

[low shard]

yw and lmk

[unique heath]

Well, it’s between either the voice changer or doing vocal training, correction, continuing vocal training

[low shard]

Well yeah, try the program with the best settings, if it still sounds bad, it’s a model issue

tho this will solve performance issues

[unique heath]

I will definitely try it, because am all for trying out different good voice changers so I will let you know!! thank you

[hybrid ocean]

Yeah

I have RTX 4050

I want to use it in realtime calling

[pastel oak]

https://rentry.co/forkvoicechangerguide#download-for-nvidia-gpu-on-windows

[hybrid ocean]

whats this ?

[true ledge]

Yo could someone help me get the ai software for the voice changer, because might get hub looks alot different to the vids

@acoustic scarab

[simple ore]

literally 2 messages up

[true ledge]

I'm a bit dyslexic so I was trying to ask if someone could like call and help

Because I was trying to get it for a video

@simple ore do I install the 5000 series one is i got an rtx4070super because that's the only one I see for windows

Or the download for nvideo gpu and windows

[simple ore]

<5000 series - you download the regular

[craggy bough]

read what it says

"NVIDIA RTX-5000 series, the newest release of GPU's, require a separate download. You do not need it if you have an older GPU, follow the normal Nvidia link in that case

https://github.com/IllIlIlIllIl/voice-changer/releases/tag/b2335"

[simple ore]

5000 series - you download 5000 series

how hard is it?

[true ledge]

It's a bit hard when I'm dyslexic

I read stuff then I read another and then it gets confusing

Alr appreciate the help btw

[knotty moth]

https://tenor.com/view/frye-dyslexic-gif-4978104

[low shard]

-realtime

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

1st link Wokada deiteris fork

[karmic bronze]

do you know any way to not have like the robotic like sound? im spanish and almost all models in spanish sounds robotic, if i speak in english they sound better

[low shard]

show a screenshot of ur wokada

[nocturne patrol]

guys will people hear me like how i sound in discord mic test or no i’m really confused

because i test it in voice changer and the discord mic test both with the voice changer working and it sounds more realistic and cleaner in discord mic test

[low shard]

show a screenshot of ur wokada and discord settings

[nocturne patrol]

i’m not home i can do it when i’m home

[viscid moss]

@ruby basalt

[patent trellis]

💻 Local Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[viscid moss]

First link

pls follow the steps described there

[nocturne patrol]

???

[echo schooner]

https://huggingface.co/wok000/vcclient000/tree/main
which one do i download? i just reset my pc and cant remember

[wary oxide]

hi guys, im trying to train a voice model but it keeps saying no files found even when i uploaded like a 5 minute sample of the og

can anyone help?

oh its disabled

apparently

[low shard]

Your settings are good, If it sounds bad it's just because of the model

[low shard]

You don't, this is the original wokada, which is worse from quality compared to the new Wokada deiteris fork, video tutorials are outdated since a while

What's your PC GPU?

[nocturne patrol]

oh okay, so will people hear me like they do in discord mic test and what do i put my discord output and input to so i can use the voice changer in call

[low shard]

yes, set discord input as line 1 and output as ur headphones

[nocturne patrol]

what abt my mic?

im using my microphone

[low shard]

A VAC (Virtual Audio Cable) makes a fake audio device, used to re-route the audio of different programs

In Wokada context, it's used to get the output of wokada as the input in other programs

so line 1 takes the wokada output and puts it as input in discord

[nocturne patrol]

ohhh

[torpid haven]

why when i'm using vcclient the voice lags so much? i have a rx 6700 xt and i followed this guide

[real hemlock]

where can i train voice model

[brittle wing]

-colab

[patent trellis]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[cloud finch]

Should I get the Applio app or just stay on the one I already have

[crystal gull]

It'd be nice if you gave info like, what you already have, what your usecase is, yada.

[pastel oak]

If youre replying to the forkvoicechangerguide link and saying "vcclient" then you definitely didnt follow the guide, those are 2 different things
Download the amd version from the link that i sent and make sure to do everything it says

Especially changing gpu to cpu to gpu after changing chunk or extra

Use recommended chunk and extra too

[hallow thistle]

I don't know how you followed the guide. But the "fork" W-Okada DirectML should supposedly look like this, not the one with too many files in one folder.

Also, make sure you didn't follow any tutorial video on YouTube before this one.

[hallow thistle]

Please specify on which reason you should stay for Applio.

[hallow thistle]

This link is the Hugging Face repo for original W-Okada versions, which are all outdated. What is your PC GPU? There's a better one to download.

[elfin nebula]

How do you get a more accurate voice model with fish speech? I don't understand the guide in fish speech. Which words do I use for it to be more efficient. I was planning to just slam a 2 hour recording. But I don't think it'll be necessary

[hallow thistle]

You have a model maker role, so I think you know where to train voice model.

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[hallow thistle]

Which RVC program/website are you trying to use? Is it Weights? Because Weights often limits most features for free users.

[elfin nebula]

If it's me you're asking, it's Fish Speech by Zluda

[hallow thistle]

I wasn't asking you. I was asking another user.

[glossy falcon]

Hello guys
I was making a jjk edit i wanted the voice of the narrator from hakari voice lines
i am really really new to this
ai voice stuff
can someone help

[formal wind]

Alr so I don't know if this is a channel I can ask for help in with this topic.

So I'm trynna figure out what ai voice was used in this

I've looked through the internet and haven't found a tts like that

[gusty granite]

Hi, can anyone please give me a guide how to download the voice changer?

[gilded trail]

using the applio for training a model, I read the guide but I didn't understand if the dataset has to be a 10-30 minutes audio or several 10-15 second audios?

[low shard]

realtime for calls? What’s ur pc gpu?

[low shard]

10-30 minutes are suggested

[low shard]

@gusty granite actually let’s talk in https://discord.com/channels/1159260121998827560/1367065669862031501

[outer valley]

Just a short question.
Is deiteris' optimized W-Okada still the best in terms of performance and quality (if the model is good)?
Or is there new stuff out there (maybe paid?).
Didn't checked for a long time but was interested if there is any new bigger steps with RVC things.
I'm mostly interested if the pitch situations are still a thing or if this got better?

[pastel oak]

!give-media-perms 2h @stoic aspen

[dull iron]

Oops, We got some errors!

• [Error] Giving <@&1313370885650124821> to @stoic aspen for 2h

[pastel oak]

Make a post in #1192011222023950368

You can post pics there