✨│ai-help

[simple ore]

every important module is the same

[golden walrus]

ahhhhhhhh

that one

[analog obsidian]

https://github.com/Convbased/Convbased-Studio

[golden walrus]

lmao, i thought i need to change in their config

[simple ore]

the repo actually does not have any of this

[golden walrus]

my bad

[simple ore]

no

I said change the config in your model's folder

[analog obsidian]

pretty sus

looks legit but at the same time, it doesn't

lol

[golden walrus]

hey, i'm not the brightest bulb you know

[analog obsidian]

do even they have a demo?

[wheat copper]

do i just put the projct name, sample rate and click training

after ive loaded the backup

[analog obsidian]

this may be a virus so dont download it

[simple ore]

download the model

not the whole thing, that's unnecessary, I think

[simple ore]

when you load a backup, you provide a name of the model

then you run the cell 'set training variables' with the same name, sr

then to resume you just start training cell

[wheat copper]

im using one that gives u a ui

also for some odd reason when i clcik start training it does the index

https://tenor.com/view/fortnite-jonesy-peely-season9-gif-25623186

[simple ore]

okay, load backup then with the proper name, then start traning from ui after providing the same name and sr

[analog obsidian]

hmm ok, but honestly i dont wanna try it, the whole thing seems fishy and weird

[simple ore]

because the training stops with the missing file

[wheat copper]

i shouldnt have a missing file i did nothing to the backup file so everything should be in place

[simple ore]

again, verify that your backup has the file

and that after you load the backup the file is visible in the file browser thingie

[simple ore]

and thanking RVC-boss while using Applio code

rude

[analog obsidian]

LOL

😭

[simple ore]

and contentvec embedder instead of chinese hubert

what is it?

[viral mason]

I'll just show this, they're having an issue where the new voice changer crashes but I don't understand

[solid vortex]

can someone help me because i am having an error using 5080 series card

[simple ore]

which voice changer?

vonovox or detritus fork?

[solid vortex]

W-Okada RealTime Voice Changer Client (Fork)

[simple ore]

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

[solid vortex]

yes that version

[simple ore]

so you downloaded 3 files

and unzipped, without any errors

[solid vortex]

yes all that

and i extracl all that file

[simple ore]

and what is the error you see when it crashes?

[solid vortex]

Error
unhandlederjection
no error stack

[simple ore]

on UI or in the console window?

[solid vortex]

on web browser

[simple ore]

what is in the console

[solid vortex]

wait me a moment let me extract all the files

I will re-download the file and follow your instructions

Which program should I use to unzip these files, 7zip or WinRAR?

[solid vortex]

I have downloaded all the files, now how do I extract them properly?

[simple ore]

the file is spit into multi-parts, but 7-zip should be able to handle that

[solid vortex]

I have extracted all the files using 7zip and now I should run the MMVCServerSIO.exe file?

[simple ore]

yes

[solid vortex]

TypeError: Cannot read properties of null (reading 'modelSlots')
TypeError: Cannot read properties of null (reading 'modelSlots')
at i (http://127.0.0.1:18888/index.js:2:1249327)
at Object.updateServerSettings (http://127.0.0.1:18888/index.js:2:1249559)

[simple ore]

you need to add a model it looks like

[solid vortex]

I am getting this error now.

[solid vortex]

I added it like that but the system reported an error

The notification system is
Error
unhandledrejection
no error stack

im add the model and start Exceptions.VoiceChangerIsNotSelectedException: 'Voice Changer is not selected.'
2025-06-19 01:45:10,716 ERROR [VoiceChangerManager] 'Voice Changer is not selected.'
Traceback (most recent call last):
File "voice_changer\VoiceChangerManager.py", line 212, in change_voice
audio, vol, perf = self.vc.on_request(receivedData)
File "torch\utils_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "voice_changer\VoiceChangerV2.py", line 159, in on_request
raise VoiceChangerIsNotSelectedException("Voice Changer is not selected.")
Exceptions.VoiceChangerIsNotSelectedException: 'Voice Changer is not selected.'
2025-06-19 01:45:10,787 INFO [VoiceChangerManager] update configuration serverAudioStated: 0

[simple ore]

well, it is the usual voice changer stuff, not the 5080 issue

[solid vortex]

So where does the problem come from?

[simple ore]

show UI

[solid vortex]

where to get UI ?

[simple ore]

the main window

[solid vortex]

what does it look like

[simple ore]

looks good to me

[solid vortex]

Is it because the model is old?

[simple ore]

no

[solid vortex]

Now what to do to fix it ?

[simple ore]

someone else may help

i dont use the voice changer

[analog obsidian]

set the sr value to 48000

and use wasapi

not mme

[hazy eagle]

should i turn off my hardware acceleration?

[solid vortex]

thank you it working

[polar fern]

do you have a girl voice for me . i have a deep voice . when you have one pls dm me. thx

[brittle wing]

Hmm can I send you several checkpoints so you tell me which one is the best cause I'm unsure

Like audio samples

[simple ore]

we have qc people here

[golden walrus]

guys, i have a question, how can i enable upscaler for vonovox tho

i think last time i got issue with klm is using 32k not 48k so it doesn't give the best result

[simple ore]

it may be paid thing

[golden walrus]

it's time to pay

[brittle wing]

Hmm yes but I still don't wanna apply for model maker

[golden walrus]

it's here?

my god, i'm so dumb

[brittle wing]

You're a QC can I dm you the checkpoints & u tell me

[brittle wing]

No you're not you're just nervous and distracted

I swear.

[analog obsidian]

i think the vonovox upscaler tries to upscale the output of the model

[brittle wing]

Get some rest

[analog obsidian]

but imo thats not realy useful

most people can't tell the difference between 32k and 48k

[golden walrus]

ah ye, same stuffs

reeeeeeeeeee

time to take a rest, i guess

[brittle wing]

Can you help me

[analog obsidian]

i don't accept dms, no

[brittle wing]

I can't send them here either

Then how...

Well see at least I asked you for permission!

[analog obsidian]

hear your epochs, then keep one that doesnt sound noisy or robotic

[golden walrus]

uiiiiii, KLM works like a charm. noiceeeeeeeeeeee

[brittle wing]

With least artifacts?

And flaws?

[viral mason]

look in the voice models channel lmao

[analog obsidian]

for model submissions we only accept models that have decent quality and the least amount of artifacting possible

[brittle wing]

Okay I know

Cause you're an expert

[analog obsidian]

you don't need to be an expert to notice something is wrong with your model

they're easily audible

as long your model isn't a robotic mess or glitchy, it should be enough for a submission

we dont really ask for perfect models, just models that were properly cleaned and trained correctly, without overtraining

@low shard can u give iceicebaby media perms?

so she can send audios of her model

[flint anvil]

editing the code is so different than running the prebuilt version :/

because i need to recompile the code on the repo (which doesnt work anymore for some reason)

but gh workflows are using win-latest and i think whatever deiteris used in his workflow is outdated

[patent oriole]

hello so im making an app that will have an feature like rizzgpt. where u can ask and chat bot for advice on an answer. the prompt i should give for example chatgpt what could that be? i tried some but its just not right

[jaunty star]

apologies for the late reply (discord didn't notify me), but here's my wokada settings

usual performance is around 40-70 ms while gaming

[viral mason]

128 is crazy low

how good is it?

[jaunty star]

it's been a while since i used this actually, but it's usually 50/50

sometimes it's really good, sometimes it's pretty unusable

[viral mason]

try 360.0

see how it is compared to 128

[jaunty star]

nvm, my fork okada completely bricked for no reason

i run the application and it gets stuck on compiling hubert_base then closes

[simple ore]

delete hubert model, probably damaged/incompletely downloaded

@jaunty star

and restart

[jaunty star]

alright i'll give it a go, thanks!

[crude apex]

any idea how to stop the wokada app from echoing what I said when using RVC, like for example I test saying "hello" and it would keep repeating it until its completely distorted

[alpine flare]

• rx 580 8gb 2048sp
• girl voice on w okada
• when i try to speak, the "Res" shows high values, around 8.000-40.000, and the always the voice is stuttering, its impossible to hear the AI
• i watched several videos, i installed the MMVC server, the vc client, i used the google colaborator (it happens the same issue, and after 2 minutes using it, the server suddenly disconnect), so i cant send you a tutorial link bc i watched a lot of it and everytime is the same issue (im sorry for my english, im brazilian)

[simple ore]

fix the audio routing

[crude apex]

Is there a technical meaning for this or are you just talking about input and output? In that case I already done that and got the right channels

[simple ore]

are you using headphones?

do they bleed audio that the headset mic picks up?

or perhaps you''ve picked inputs and outputs incorrectly

[crude apex]

Using headphones yah and this might be the case. As it happened before to me, but I never figured out why it happens.

[simple ore]

assuming you have a simple setup mic -> voice changer -> vac -> discord, there's no other source for the echo

lower mic sensitivity or lower headphone volume, see if that solves the problem

[viscid moss]

@iron flame

-rt

[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

[iron flame]

whats rt

oh

what do i download

[viscid moss]

it depends on ur hardware

[iron flame]

k ima just see if i can find another one cause i know nothing about computers

[knotty moth]

the upscaler would add more performance load, I suppose

and ofc not like the graphic upscaling but more of like supersampling (rendering in twice the target resolution)

[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

[arctic sand]

audio client or server which is best??

i have audio interface

[neon cosmos]

@mods can somone give me streaming perms please

[arctic sand]

I have an audio interface. Do I need to install flexasio? Also, which one should I use, server or client? My audio interface is Scarlet Solo 4th

@low shard helppp

[pastel oak]

Use server mode with Windows WASAPI for best audio (cant use noise suppression in this mode)
flexasio is optional, can provide less delay but not noticably much, maybe 5-10ms

[arctic sand]

Do I use wasapi even if I use the audio interface? Also is it best audio quality?

not a client mode?

[pastel oak]

Best audio quality

client uses MME by default which is old

wasapi is more up to date, less delay, more stable

There is no correlation between your scarlet audio interface and what audio mode you use on the voice changer

[arctic sand]

So I can use wasapi in server mode without having to use flexasio, right?

[pastel oak]

Yes. You choose for either one of them, both are essentially the same for the normal ear

Wasapi requires no setup, flexasio requires downloading extra stuff, you probably have seen the guide for it

Youre fine with wasapi, make sure to select windows wasapi as the [prefix] on everything. sr 48000

[native nexus]

Message me privately for more information

[pastel oak]

Bro is ominous

hes not in the server anyway

[reef zephyr]

Hello Guys
What is the best and most natural-sounding human voice I'm looking for one that is natural and has been trained a lot.

[echo rune]

hi guys! why i cant rename rvc model in realtime voice changer (fork)?

it instantly crashes

Looks like there's a bit of a problem.
unknwon message

If you clear the information being managed by this app, it may be recoverable.

[latent kettle]

Firstly he left the server
2nd tell him here in server. You are allowed to do that

[vocal glade]

anyone has any solution to my problem everytime i use w okada the voice is delayed so much and sometimes it doesnt register my voice at all? i dont have good pc if that matters

[knotty moth]

the bare minimum gpu to get decent performance is RTX 2060 (or RX 580) or 3060 as recommended

to post the voice changer screenshot, pls tell me first

[vocal glade]

how do i check these specs im not very a techy person myself

[vocal glade]

if i tell u my cpu can u tell me if its able to run it?

[latent kettle]

Just go to task bar and right click and open task manager check for gpu and send the image

[low shard]

Set chunk to 300ms
You can also try other models and server mode for less delay

[low shard]

all video tutorials are outdated, don't follow any

WOKADA is banned on Google colab as it uses a Web UI, it can't be used in the free tier, you have to buy colab

Also, Colab is a cloud computing service, meant for remote good PC for people with a bad one.

Your PC is the bare minimum to run WOKADA, are you going to use it in games? Or just discord VC?

[low shard]

Btw is it fixed? If not, send a screenshot of ur wokada

[low shard]

You can't edit the model name directly on the WOKADA deiteris fork web UI

You have to rename the model files and then reupload the slot

[low shard]

Elaborate:

• your PC GPU
• what you want to do
• the tutorial link you followed
• a screenshot of your wokada

[low shard]

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

[low shard]

CPU doesn't matter at all, you need a good GPU

[arctic sand]

thx

[shut ember]

help me, how do I use the f0 curve feature for TTS in Applio? i’ve tried extracting it from the 'extra' tab and got a .txt file, but when I tried to load it, a cuda error appeared. i also tried changing the extension to .csv, but it had no effect

[rotund zephyr]

guys is there any real time voice changer that works? wokada for me doesnt work cuz im a free user

[crude flame]

W-okada

It's free so IDK what you mean by "it doesn't work cuz I'm a free user"

[hallow thistle]

CPU (central processing unit) and GPU (graphics processing unit) are two different units in computer. W-Okada would work with any CPU, but running W-Okada with only CPU is not recommended. A good and decent GPU is highly recommended for W-Okada.

To check your PC CPU and also GPU, open Task Manager or press Ctrl + Alt + Esc, go to Performance tab, spot where CPU, GPU 0 and GPU 1 are in the left panel, and click either one of these to reveal its full name on right panel.

[hallow thistle]

W-Okada is a free and open source program. If you use an outdated version of W-Okada, it would sure not work.

[rugged hill]

Where/How do I train ai models for W-Okada?

[rotund zephyr]

Wokada's Deiteris' Fork WARNING: This colab works, but even with the recently added encryption, it gets free tier users flagged for using a web ui

i meant this

[hallow thistle]

W-Okada does not have a feature to train a voice model. It's mainly a realtime voice changer program. There are RVC and Applio, which both can train an RVC voice model, but some online alternatives are available.

[rugged hill]

Thank you

[simple ore]

that feature does not work

[viral mason]

4 days since weights has created samples for models, rip weights

[hallow thistle]

If you say "cuz im a free user", that doesn't tell how you use Google Colab.

!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.

[shut ember]

ok bro, no wonder no one talks about it on youtube or in github issues

[rotund zephyr]

well i use it normally on an alt account but still got flagged a few seconds after run the first cell

[simple ore]

i think the original idea was like auto-tuning or fixing some parts of the audio where f0 is too high, but aint nobody got time for that

[hallow thistle]

Deitris fork W-Okada is banned on Google Colab for free users. The original W-Okada is still broken. Use Kaggle as an alternative instead, but you'll have to register with your phone number on this one.

[shut ember]

ok bro, thank you for your answer

[rotund zephyr]

ok thanks for this

[rotund zephyr]

is there a link tho?

[low shard]

you shouldnt really look into youtube tuts for rvc btw

[low shard]

what's ur pc gpu?

colab is a cloud service that lets people with a bad pc use good pc, but its limited for free ofcourse

you shouldnt use it if you got a good pc

[rotund zephyr]

nothing really good, iGPU amd athlon 200ge thats why i go cloud

[shut ember]

that was just my alternative, because as I said, no one discusses that topic in the github issues either

[low shard]

yep your cooked, your only options are:

• buy a better pc
• use kaggle: another cloud computing service, gives 30 hours weekly, needs a phone number verification, and is harder to use
• pay for colab: colab doesnt allow at all web uis in the free tier, and you just risk to get disconnected before even actually using it, plus you can get your account banned from using colab

you can't expect to use ai on bad hardware, nor use cloud for free forever

soo, what's your choice?

[rotund zephyr]

kaggle is good

[low shard]

it would be better to also check https://docs.aihub.gg and https://docs.applio.org instead, youtube shouldn't be your option at all

[shut ember]

kaggle + email + cheap online otp number

[rotund zephyr]

ill just look up the guide for it

[shut ember]

ok bro, thank you for your answer

[brittle wing]

is rx 580 cooked too?

[hallow thistle]

Folow this guide. https://docs.aihub.gg/rvc-voice-changer/cloud/w-okada-kaggle/

[low shard]

dont use video tutorials

guide for kaggle and colab

I have warned you about Colab, it's better you use Kaggle

[low shard]

what do you want to do?

[low shard]

yw

[brittle wing]

answer

[low shard]

i cant if you dont tell me what you want to do

[brittle wing]

its a questionb

[low shard]

this is a general ai server

we dont use only a single program

[brittle wing]

for ai voice

[low shard]

elaborate

[rotund zephyr]

ok thanks bro

[low shard]

pre-recorded audios? making models? realtime for calls? realtime for games? @brittle wing

[brittle wing]

realtime for games

[low shard]

there's different audios, and the answer would be different

[brittle wing]

mb thoguht this was for one program

[low shard]

which specifically?

[low shard]

nope, this isn't an ai voice server, we are trying to do other things too

[brittle wing]

w-okada

[low shard]

i meant which game?

[brittle wing]

roblox

[hallow thistle]

If you're looking for W-Okada the realtime voice changer, it would work on AMD Radeon RX 580, but that's it, it won't be really that fast.

[low shard]

ehhhhh

you may wanna try with wokada deiteris fork, which is the best version https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/

locally you will have unlimited time

your pc gpu is bare minimum, but worth a try

just remember to play on LOWEST SETTINGS

else, your other option is cloud (remote good pc), but it has limited free time as you can't expect everything to be free forever

or the best other option would be just buying a better pc

@brittle wing ofc let me know

[brittle wing]

its lagging when running on its own

[hallow thistle]

If you feel like W-Okada is not fast enough on your old GPU or just lags a lot whenever you try to make it work, you can try Kaggle instead.

[low shard]

lemme guess, you're using a video tutorial from yt?

[brittle wing]

yep

[low shard]

all video tutorials are outdated and use an over year old version of original wokada

[low shard]

then thats why

delete everything you got off youtube

[low shard]

@brittle wing check this instead

only written guides are up to date

[brittle wing]

ok thx

[rotund zephyr]

can confirm, im using kaggle rn its so good guys

[low shard]

be sure to read it carefully, and lmk for issues

[hallow thistle]

The original version of W-Okada is old and outdated; its DirectML variant is also bugged too. Try the better one.

[rotund zephyr]

well nvm i got this for some reason @hallow thistle@low shard are u guys able to help?

[low shard]

can you elaborate what link are you using and what did u do exactly?

[rotund zephyr]

well i used Public W-okada Voice Changer for kaggle and i clicked on some prebuilt models then it just did that

[low shard]

Public W-okada Voice Changer for kaggle
send the link

I feel like you arent using the one I told you

also there shouldn't be any prebuilt models

[rotund zephyr]

yeaaaa man my bad i clicked on the wrong one the one i should be on is just called voice changer public

is that right?

[low shard]

did you read https://docs.aihub.gg/rvc-voice-changer/cloud/w-okada-kaggle/ ? it should be the one from suneko

[formal badge]

Hey guys, quick question... I cloned the github for Wan, and downloaded all the safetensors files and the two .PTH files (along with everything else when github clones a repo) and it's about 78gb...

I also followed a guide for Wan on youtube because I believe I don't need the full version since I'm only using an RTX 3060, so I just want to be sure there's nothing I need from the github repos (when it comes to the safetensors or .PTH files) as I have the "cut down" version...

Just want to confirm I can go ahead and delete those to free up some space since I have the "smaller" version...

https://imgur.com/a/8HGfqJf (edited to show a screenshot with size of files)

[rotund zephyr]

yes i did but i saw this on the "your work" section on kaggle and thought they had no difference

so i just clicked what seemed more legit

[low shard]

nope, those are completely different

the 2nd is the modified hina version of original wokada, which is broken and not maintained

the 1st is from suneko (aka deiteris), which is the updated wokada deiteris fork

[rotund zephyr]

i havent used rvc's in a while so yeah i didnt know

[rotund zephyr]

well im using the one from the guide now

will update u if it works or not

it should tho

[viral mason]

the heck is this

[low shard]

this isn't rvc btw, it's wokada

RVC = Retrieval-based-Voice-Conversion, the best Few Shots Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models. Technically, Mainline RVC does have a go-realtime.bat (aka RVC-GUI), but it's pretty messy and outdated so it's extremely not suggested for realtime.

Wokada = uses RVC for realtime inference. There's 2 main versions, Original made by Wok, and the most suggested one is Deiteris Fork (modified version)

but yeah dw, i just explained you above in case you want to know the right terms

[low shard]

alr, lmk

[rotund zephyr]

works fine, thanks, sorry for the misunderstanding on my part lol

[brittle wing]

its supposed to be on web only?

[low shard]

yw, did you also install vac lite? (and not vb audio cable)

[low shard]

it's just the web user interface, it still runs locally, check https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window for more info

[brittle wing]

oh it actually lags less now

[low shard]

show me a screenshot of ur wokada so i can suggest u settings

[brittle wing]

my main problem is with muffle a tthe end

[viral mason]

is 512 better than 360?

[brittle wing]

chunk?

[viral mason]

yea

[brittle wing]

An error occurred during voice conversion. Check command line window for more details.

[low shard]

uninstall vb audio cable

it can cause issues on windows

read and get vac lite from https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

set monitor to ur headphones to hear urself optionally

[viral mason]

I just use voicemod to hear myself

much better

for me at least

[low shard]

wouldnt that cause just more delay? its going through another application too

its suggested to use just the monitor option with ur headphones

[viral mason]

I only have like a 2 sec delay

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay @brittle wing u can optionally follow this part to get even less delay btw

[viral mason]

and can add decent effects to the okdada voice

[rotund zephyr]

may i ask what kinds of issues tho?

does it affect the voice or anything? or just delays?

[low shard]

randomly stop working, already many many users reported that when we used to suggest it time ago

tmr it can just stop working at all

or the day after tmr

or in a week

who knows

vac lite is way easier and less bugs

[rotund zephyr]

oh aight

[low shard]

unless you like playing russian roulette

😭

[low shard]

you were using vb audio cable too? its better to use vac lite

i can also suggest you the best settings if u send a screenshot

[viral mason]

what's the difference between vac lite and vb cable?

[low shard]

I explained it above

[viral mason]

ah

I never had issues with it ever

[rotund zephyr]

me and satical have the same problem with the muffle at the end but i think thats the models fault correct me if im wrong tho

[low shard]

many other issues reported otherwise, its much more suggested to use vac lite for a easier life

[low shard]

are yall talking in dms? he didnt say that

btw, wdym with muffle? like cutoff?

[viral mason]

will it improve with less delay?

[rotund zephyr]

@low shard

[low shard]

not sure about delay, but its better for being more safe and easier to use

other people like @pastel oak can confirm you too that vac lite is a much better pick

[low shard]

@brittle wing do yall mean like a cutoff at the end of ur words?

[viral mason]

I think u should provide a video example with audio

[brittle wing]

nah turns out it was just a discor dprobnlem

all good now thanks

[rotund zephyr]

idk bruh it just sounds like its glitching out for a milisecond, might just be background noise im thinkin

[brittle wing]

it also uses significantly less cpu

[low shard]

did you get vac lite?

also you may wanna check https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay for even less delay,

or turn on force fp32 mode on in advanced settings for a bit more delay but bit more quality

[viral mason]

should def send the vac lite download link

[low shard]

it's in the guide already

[low shard]

may you screen record to let me hear? also you can try echo/noise suppression (sup2)

[rotund zephyr]

what does vac lite appear as? is it line 1 virtual audio cable?

[rotund zephyr]

yeah was just background noise mb

[low shard]

Yeah, in output of WOKADA it should appear as Line 1

[viral mason]

I tried using it with voicemod and it doesn't even work but I could hear myself for some reason

without monitor

[unreal ledge]

hello when i try to use any voice on the voice change it just does not work. yesterday it waotked foe me but now suddnly it does not

[brittle wing]

i did

also my ai voice audio bleeded into every track on obs, does anyone know how to fix that

i put diferent mics and i wasnt listening to the voice while recording yet it still appeared in every track

[unreal ledge]

and the ai voice tune does not go to zero when i swiitch to another voice what should i do

[brittle wing]

I'm gonna use mdx23c server for my inference vocals cause anyway dereverb is way too aggressive and removes half of the words, making my model sound bad, like gibberish

[low shard]

elaborate

[low shard]

wdym bleeds?

[last cobalt]

what should i use if i want real time changer to use in games? vonovox or the wokada fork

[low shard]

what's ur pc gpu?

[last cobalt]

3070

[low shard]

wokada deiteris fork is the most used one generally

vonovox is a more recent project that is nvidia only, @crude flame tested it and said it might have a little less of delay but i didnt test it myself

[last cobalt]

alr thanks

ill try vonovox

[azure patio]

how to use uvr deecho?

[tribal salmon]

what are the best pitch,formant and index settings to sound like a woman (i am a man) with wokada fork?

[viral mason]

just pitch it up until it sounds right

1 at a time, formant isn't really needed

[tribal salmon]

thx will try

[muted coral]

so im confused

currently running 2 gpu's and for some reason the client will not work on my second gpu

[viral mason]

why are u running more than one 😭

[muted coral]

Load assignments

But honestly gotta get a new mobo anyway cause i just found out my shit fkin disables my m.2_3 so like i dont fkin have that xD

The pcie 2 slot shares lanes with my m.2_3 so its either one or the other

I bought a x870 eagle so thatll fix my problem

Should be here tomorrow

[simple ore]

are you for real?

x870 eagle has one x16 PCIe

the other runs at Gen3 x1 speed

with x870 at best you can get 1x Gen5 x16 + 1x Gen4 x4

with x870e you can get 1x Gen5 x16 + 1x Gen5 x8

[knotty moth]

voice changer? why do you want running dual gpu for that?

you will never enjoy realtime performance benefit from it

[knotty moth]

there's nothing wrong on doing it, but getting a faster gpu is more likely to solve the performance issue

I suppose the performance hit on the pcie shouldn't be noticable enough, unlike in this case: https://www.youtube.com/watch?v=kEsSUPuvHI4

[fading barn]

is UVR still the way to go for isolating speech when cleaning up an audio file?

[median monolith]

Does anyone know or remember the name of that one YouTube channel that mostly consists of AI related news, where the avatar of the channel owner is like a little cloud head guy with a blue hoodie? It was something like the image below.
Trying to find it again because I found it really interesting when I watched a long time ago.

[inner pivot]

https://youtube.com/@bycloudai?si=YB8-ij8uoQhD6UnJ

Perhaps this is whom you're looking for

[simple ore]

for x16 maybe not, but using gen4 x16 card in gen3x1 slot is yikes

[median monolith]

Les go! It is! Thanks a lot 🙏

[brittle wing]

I can't find the "apply for model maker section"

[knotty moth]

type a hashtag and you'll find it

[brittle wing]

I know but I don't wanna be spammin'

#outdated-model-maker-role

[opal flint]

what to do if ai voice sounds very glitchy

[brittle wing]

How long should I wait for an answer.

[brittle wing]

Use a better model

[opal flint]

ive used like 4 already but sometimes they sound very glitchy

i wanted to know if there was maybe a setting I missed

[compact patrol]

hello, I can't find the saneking model

does anyone have the model left?

[fresh pilot]

is the ai voice changer i just watched a youtube video on it and it lookedd cool but i keep find it

[slim schooner]

Hey guys, so i managed to clone a voice. However, it still sounds a bit robotic at very random times. I trained it up to 150 epochs but i feel like i should train it a bit more. My solutions for a better clone are 3 things: more training time, selecting a better pretrained model, and selecting the right embedder model.

I'd like to note that the current cloned voice sounds good but the robotic effect is what really degrades it. To solve the robotic effect issue should I do all 3 things I mentioned above? It's a 50 min audio data set that I'm working with. The audio is cleaned and free from any background noise.

I'd also like to note that I'm cloning a voice to speak in Spanish. What embedder and pretrained model should I use in this case? I used TITAN pretrained model which came out good so I may keep that. Should I also choose a custom embedder model?

Any help is appreciated!

[muted coral]

So for some reason running some games or whatever that puts a heavy load on the gpu will start cutting out the ai making it unusable. So the thought of splitting that load to having one gpu run certain task and the primary just being used for everything else itll kinda help reduce the risk of shit not working

[low shard]

elaborate:

• ur pc gpu
• what tut link are u using
• what u want to do
• what model link are u using

[low shard]

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:

• make it yourself with our docs guides

[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

[low shard]

dont watch youtube tutorials for rvc and wokada, they are old

what's ur pc gpu? what do you want to do?

[fresh pilot]

urm i dunno my pc specs but my computer is good and i just wanted it for when i play games

@low shard

[low shard]

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

You must know the GPU to use the program

[low shard]

no need to ping

[low shard]

It's not a 1 click program, you will have to find the settings and version depending on your GPU

and you have to check if your gpu is even good enough

[low shard]

I'm guessing you want a realtime voice changer for games

if you tell me your gpu i can redirect you to the written guide you can read for getting it and help you out

[fresh pilot]

i have a 4060 ti

[fresh pilot]

yea

[low shard]

good enough, what game? be sure to play on lowest settings while using the program

[fresh pilot]

cs2

[low shard]

alright

-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]

read up the 1st link

wokada deiteris fork

[fresh pilot]

ok thanks

[low shard]

yw and lmk

[simple ore]

limit fps in the game

you dont want CS2 to run 400 fps while competing with the voice changer

[short oracle]

hello, idk why its happening but my voice changer doesnt work at all and the console is showing me this

2025-06-20 13:02:16.9893382 [W:onnxruntime:Default, scatter_nd.h:51 onnxruntime::cuda::ScatterNDWithAtomicReduction::ScatterNDWithAtomicReduction] ScatterND with reduction=='none' only guarantees to be correct if indices are not duplicated.

[versed anvil]

the model i wanted is deleted sobs

[knotty moth]

set fps limit to your monitor refresh rate or at most 120/240

[brittle wing]

ai voice appears in every track even tho i seperated them

[spark shard]

Hey, i need a bit of help i downloaded the client and also installed the virtual cable but my mic isnt being recognised what do i need to change?

[low shard]

He left

[low shard]

Elaborate:

• ur PC GPU
• what u want to do
• what tutorial link are u using

[low shard]

Client, in computing, means a desktop computer that can obtain information from a server

We are a general ai server, we don't do only a specific field, so your help request needs to be a little more detailed

Elaborate:

• your PC GPU
• what you want to do
• what did you download
• what tutorial link are you following
• a screenshot of the issue

[short oracle]

5080 laptop, i want the voice changer to work and this one

https://www.reddit.com/r/TransVoicechangers/comments/156y1g0/intro_to_wokada_voice_changer/

[low shard]

that is an 2 years old video

ofcourse it doesn't work anymore

the version used is more than ancient

ai develops at sonic speed

it doesn't even support rtx 50 serie

[spark shard]

I have the RTX 2060 and want this voicechanger to work ^^ https://www.youtube.com/watch?v=SxdnGxicJOg

[low shard]

forget everything you see from video tutorials, they use an old version of original wokada

[low shard]

also im guessing you mean realtime for calls

if so, uninstall everything you want from videos

-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]

read the 1st link, wokada deiteris fork

theres no updated video

[low shard]

that one wont ever work

it uses an over year olod version of original wokada

never trust video tutorials for voice changers

[spark shard]

Does the wokada work right now

[low shard]

the one you got absolutely not, you need the latest wokada deiteris fork for realtime voice changing

forget everything about the video

uninstall everything

vb audio cable too, it causes issues on windows

-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]

read the 1st link, wokada deiteris fork

the only way to get the updated one is by reading, trusting random video tuts will just cause you more issues basically

so please read it carefully

[spark shard]

Thanks for the help!

[short oracle]

Yeah

[low shard]

yw and lmk

[low shard]

alr, follow what i said and lmk

[short oracle]

It will

Worl on my laptop?

And thankk youu

[spark shard]

I cant do it right now but will do it later today and update you ^^

[low shard]

yeah your laptop is good

all u gotta do is read up the guide and let me know for any issues or trouble, i can also suggest the best settings

[short oracle]

Thank uuuu

I wanna mainly use it wheen i hangout with my friendss on vr chat

[low shard]

yeah dw it will work, when u installed it, go ingame, set lowest settings and show me a screenshot so i can suggest you the best settings

[short oracle]

I might not be able to test if out today cause

The power had died in my town

[low shard]

oh damn, goodluck and let me know when u can then

[low shard]

alright

[low shard]

could you show a screenrecord?

this is meant only for realtime voice changing in calls

not for ai covers

i dont understand what you're trying to do

[short oracle]

Can dm u?

[low shard]

No need to, you can just ping me there

so in case i will be sleepin or whatever other helpers can help too, since yk timezones

[short oracle]

What timezone u in?

Am in eu

[short oracle]

@low shard

[simple ore]

show UI

[minor linden]

is someone willing to clone a voice for me (for realtime voice changer) i tried myself but im stupid

[spark shard]

@low shard i cant get the voicechanger to work i installed it but as soon as i use the vaclite as input on discord it wont pic up audio

[short oracle]

i managed to get it working

but now i dont know what the settings should be

[low shard]

me too

[low shard]

u can send a screenshot and i can help

oh nvm u showed it already

[low shard]

u can optionally set force fp32 mode on in advanced settings for a bit more quality at the cost of some delay

also, you can try:
Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms)

and https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay for less delay

[low shard]

try the optional things i told you if u want, then close useless programs in background, go in game with lowest settings, then show a screenshot of ur wokada again

[low shard]

We don't allow and don't endorse model requests, either find already made models or make them yourself. Be aware of scams if someone tries to dm you saying that they will do the model for you if you pay them

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:

• make it yourself with our docs guides

[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

[low shard]

show a screenshot of ur wokada, and discord settings

[short oracle]

okay i fixed it just dont know if it sounds good

[spark shard]

i cant send pictures in this channel

[low shard]

you can set monitor to ur headphones to hear urself optionally

did u try the things i told you?

also show me a screenshot while ur ingame with lowest settings, so i can tell you what chunk to put

[low shard]

!give-media-perms 1h @spark shard

!give-media-perms 1h @short oracle

[simple ore]

use push to talk

[low shard]

crepe sucks, especially tiny, and onnx is meant for amd

[short oracle]

i figured out the chunks

[low shard]

use rmvpe without onnx

[low shard]

did u follow this part of the guide?

[low shard]

be sure the chunk is higher than the perf value, also do you need something else or all solved?

[short oracle]

uhh where do i find good voices?

[low shard]

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)
• Suggested Models for Realtime Voice Changing (Wokada)
• 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:

• make it yourself with our docs guides
• Be aware that we don't allow any model request, so don't fall for any "pay me 20 dollars and i will make the model for you" dm

[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

[spark shard]

@low shard IT WORKS NOW Thank you so so much you are the best!!

[low shard]

you're welcome

on wokada, you can optionally:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Reduce the delay via: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

do you need any other help?

[spark shard]

Oh yes i just noticed i can hear myself talking and i cant find the setting to get rid of it

[serene geode]

what does this mean? ( fcpe ; for chunk check the perf number and add 50 to it ; 1.0s extra )

[low shard]

its the monitor setting, used to hear urself and monitor how u sound like

set it to none to make it stop, or to ur headphones to hear urself

[serene geode]

what does index mean

[low shard]

you need to check the perf value at the top left and put ur chunk to that value + 50

for better help, please elaborate:

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

[serene geode]

im on mac m1 max and im using this: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window i want to make it less laggy

[serene geode]

it spikes my perf value

[low shard]

im guessing you want realtime voice changer, be aware that mac isn't that good for ai

you're using wokada deiteris fork which is the best as of rn, close every single other program in background, are u using it in game? if so, go in game, set graphics to lowest, and show a screenshot of ur wokada settings

[serene geode]

yes im using it in game and where do i send the image i dont got image perms

[low shard]

in rvc context:

• pth files: contain the voice
• added index files: contain the accent
• metadata.sjon file: it's just some extra info about the model download link if you downloaded it off weights.com, it's not needed and won't impact the actual model at all

!give-media-perms 1h @serene geode

[low shard]

what game? and set every single setting of the game to the lowest

[serene geode]

roblox and i did hold on lemme send the setings

[low shard]

set f0 as fcpe

[serene geode]

f0 det?

[low shard]

close everything in the background

[serene geode]

i did expect discord and it doesnt let me eddit f0 det

[low shard]

yes, it will be kinda worse quality but better delay

[low shard]

click stop first to edit the settings

[serene geode]

oh alr

d

did it

now what?

[low shard]

it's a dropdown menu, you need to click it and put fcpe

[serene geode]

i did

whats next?

[low shard]

put chunk to 320,

[serene geode]

is there a way u can type the value or no

[spark shard]

this just makes more delay for me

[low shard]

you can optionally set force fp 32 mode on in advanced settings for a bit better quality but a bit more delay, it's your choice

[low shard]

no, just put it smt close to it dw

[serene geode]

alr 325 it is

[serene geode]

kk

is that all the settings?

[low shard]

yeah you can't do much about it, macs aren't good for ai and you can't use wasapi or flexasio for less delay since you they are only for windows

[serene geode]

do i set my mic as my default mic or what?

[serene geode]

alr

[low shard]

you need to set
input: your microphone
output: your virtual audio cable
monitor: optional to hear urself, ur headphones

you already installed a vac right?

[serene geode]

yea

vb cable

[low shard]

you dont seem to be utilizing it

[serene geode]

oh thats my issue

alr

[low shard]

then after u do this

set ur roblox input to the vb audio cable

[serene geode]

alr

[low shard]

what? u sure u did it correctly?

[low shard]

do you need any other help?

[serene geode]

nah im good for now

[low shard]

alr lmk

[serene geode]

wait

in settings

do i put my input as vb cable

or

on my mac sesttings

for sound

[low shard]

you dont have to touch your mac settings at all

[serene geode]

i js keep my input the same for mac settings or vb cable?

[low shard]

WOKADA:
input: your microphone
output: your virtual audio cable
monitor: optional to hear urself, ur headphones

GAME:
input: your virtual audio cable
output: your headphones

[spark shard]

Almost 100% but im pretty slow in the head so maybe no haha

[serene geode]

kk

[low shard]

show a screenshot of ur wokada and ur microphone properties

[low shard]

yeah you didnt do the wokada part yet

at all

https://rentry.co/LessDelayWasapi#on-w-okada

you have to change wokada settings as said in the guide

[spark shard]

like this

[low shard]

input and output should be windows wasapi

[spark shard]

like this?

i get this now when starting

[analog obsidian]

set monitor to wasapi as well

[spark shard]

I cant hear myself this way

[low shard]

input should be windows wasapi your microphone
output windows wasapi line 1
monitor windows wasapi ur headphones

try now and lmk

[spark shard]

nope still cant hear myself

[low shard]

show a new screenshot

[low shard]

can you show like the full names in the selection menu

[low shard]

uninstall vb audio cable from your app settings to not cause any issues

[low shard]

the one selected here should be your output

[spark shard]

wait wich one do i uninstall im kinda confused now

[low shard]

uninstall vb audio cable
keep vac lite (line 1)

[spark shard]

so this one delete?

[analog obsidian]

monitor device should be the same device you have as output device in your pc

[spark shard]

jup thats the one i have selected

[analog obsidian]

you have set your input device as your real microphone?

in w-okada

line 1 should be the output, not the input

[spark shard]

it instantly switches to this

when i try to switch it to my mic it just switches back

[analog obsidian]

weird, yeah, do what nick said, uninstall vb audio cable

[spark shard]

is it that one ?

[analog obsidian]

yes this one

[spark shard]

perfect thanks

i need to reboot brb

YESSS i can hear myself again!!!

[analog obsidian]

[spark shard]

Bro i got bad news...

it stopped working again

[analog obsidian]

try a different browser

[spark shard]

Do i need to set the new browser as standart or can i just copy the url and paste it in the other browser

[analog obsidian]

first try to just copy paste the link in the browser

[spark shard]

YES

YOU DID IT

(sorry for caps but this felt like giving birth to a pc omg i love you two so much)

[analog obsidian]

[verbal mountain]

how can i fix it repeating itself a few times

[delicate kettle]

Anyone know how I can make ai covers

[brittle wing]

Hi, I submitted for model maker

[brittle wing]

It's easy but you need some vocal presets

[delicate kettle]

I know there’s sites like jammable and all the others but the sites lately been meh and I didn’t wanna pay for it

So how do I go about the vocal presets

[verbal mountain]

it doesnt just repeat what it says but stuff from my desktop aswell

[low shard]

elaborate:

• ur pc gpu
• what u want to do
• what tutorial link are you using
• wdym
• a screenshot of the issue

We are trying to be a general AI server, we don't do only voice

[low shard]

what's ur pc gpu?

[delicate kettle]

2080 super

[low shard]

Your Nvidia GPU is good enough to do inference (use models) locally (on ur pc), not the best to train (make models) even if still possible

You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides):
  • Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
  • Mainline: The original RVC
• Cloud (remote good pc, easier and faster than ur PC but it's limited):
  • Ilaria RVC Zero: fastest and simplest that you can get for free
  • Weights.com: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Applio UI Colab: max 4 hours daily, not granted, of GPU
  • RVC-AI-Cover-Maker-U Colab: Automatically separates the vocals and instrumentals, converts the voice and mix all together back

Easiest possible (automatically separates vocals & instrumentals) : weights.com & rvc-ai-cover-maker-ui
easiest cloud: Ilaria rvc zero
easiest local: Applio

[delicate kettle]

I don’t even know how to use any of those

[verbal mountain]

4070 ti
use an AI voice
none
when i speak it repeats itself depending on how loud i speak it will do it more, so say i spoke then the AI spoke the AI would repeat what the AI just said and also it says stuff from my desktop so say im watching a yt vid when the person speaks the AI then speaks saying what they said
i cant screenshot it really because its just audio

its confusing

[spark shard]

@low shard is 300ms delay bad or is that good?

[brittle wing]

You use them in bandlab but yeah you can send me a friend request so I send them to you

[low shard]

you have to read the guides I sent you

[low shard]

4070 ti
nice
use an AI voice
how? realtime for calls? pre-recorded audios?
none
what program download link are u using

i cant screenshot it
screenshot the program settings you're using, do a screenrecord too

!give-media-perms 2h @verbal mountain

[low shard]

could u show a screenshot of ur wokada

a full one with everything

[spark shard]

@low shard As sson as i go into the red perf the quality goes SO bad

[low shard]

show everything, also the above

I have a feeling you're using the wrong version

[low shard]

oh.. you're not using the nvidia version

you're using the amd/intel/cpu version..

u didn't show this part before

[spark shard]

wait so wich one do i need to download

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-nvidia-on-windows this

the nvidia one

you already enabled wasapi so you will just need to delete the one you got, download the nvidia one and use the same settings

it should work better

[spark shard]

so delete this one and just download the other one?

[low shard]

yes

please read more carefully the guide next time

[spark shard]

yes sorry

[simple ore]

if you're using speakers, using headphones

if you're using headphones, you need to fix the sound bleed that your mic catches

lower mic sensitivity / headphone volume

[golden walrus]

sir, do you know what is the cause of my playback is different from what people hear?

they said i sound like i have poor quality mic even tho the playback is pretty smooth

[low shard]

hows it going

[low shard]

elaborate:

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

[woeful adder]

yeah

[viscid moss]

@woeful adder

[woeful adder]

yup yeah

[viscid moss]

-rt

[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

[spark shard]

Somehow im still downloading i have a browser issue where whenever i download smth over 400mb it just keeps pausing

[low shard]

maybe try another browser or use a download manager

what browser are u using and did u try a speed test

[spark shard]

I finished the download now but the .exe wont start

[woeful adder]

i may be even too stupid for a guide.. i download the "Download NVIDIA RTX 5000-series on Windows
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. nvidia-5000-Series (click here to download)" thing right? i use windows

[low shard]

what's your pc gpu?

[low shard]

you want realtime voice changer for calls right?

[woeful adder]

nvdia geforce rtx 2060

[woeful adder]

yea

[low shard]

download the normal nvidia download, rtx 50 serie version is only for rtx 50 serie lol

it was said in the guide itself

[woeful adder]

w-whichone..

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-nvidia-on-windows the one above the rtx 50 serie download

[woeful adder]

aha

LOL

AHAHAHHAHHHAHAH

I FEEL SO STUPID OMFG

ĞPİKĞDWKDWPİQKDİWDİ

[low shard]

I wanted to do like that meme with a lot of circles and arrows lol

[woeful adder]

yeaa the thing u did there made me feel like a lab rat

😭

ANYWAYS THANKS

I HOPE I WILL NOT HAVE

ANY OTHER PROBLEMS

hopefully

[spark shard]

@low shard the .exe wont start now idk why

[low shard]

this is a typical error that happens when your wifi is too slow that goes even out of the time out for downloading internal model files

[spark shard]

So basically bc my wifi sucks i cant use it ?

[low shard]

you can either get a better wifi, or just follow https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#help

which will make you download internal models from somewhere else without the timeout

[low shard]

average tiktok or ig reels

[low shard]

if you do, let me know dw

[wary shoal]

lol hey does anyone know why i keep getting an error when trying to upload a custom pretrain?

[low shard]

elaborate:

• your pc gpu
• what you want to do
• what is the error
• what tut link are u using

[wary shoal]

i got an rtx 4070 im trying to create a voice model using applio and i was trying to upload a custom pretrain to the app that i downloaded from here but it just constantly shows a red error message

[low shard]

can you elaborate which custom pretrain? and how you tried to upload it?

!give-media-perms 1h @wary shoal

u can show a screenshot

[spark shard]

@low shard

like this now?

wait i had one problem on the output but i changed it

omg im going insane i cant hear myself

[woeful adder]

hahaha thanks

[golden walrus]

i found the issue. it's discord doing

[low shard]

f0: rmvpe without onnx
extra: 2.7
chunk: 120ms

on wokada, you can optionally:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)

[golden walrus]

noise suppression and automatic gain control f me up

[low shard]

any issues?

[low shard]

alr, i could also help u up for better settings if u want, be sure to not use video tutorials

[golden walrus]

i have no idea if i can improve setting further

like

[woeful adder]

extracting it rn

[golden walrus]

you know, i keep everything default

[low shard]

you shouldn't do that

!give-media-perms 1h @golden walrus

[spark shard]

Still doesnt work O.o

[low shard]

want me to check it up?

[low shard]

whats not working exactly?

[golden walrus]

so far maybe EQ band is the most confusing stuff

but i figured it out earlier

[spark shard]

I cant hear myself and it wont pic up my mic in general

[golden walrus]

[woeful adder]

yo uh

[golden walrus]

pretty neat stuffs, from dr87

[woeful adder]

is it normal to open itself at google?

i remember at the older version

it was not like that

[low shard]

did u try another browser? like u did before #✨│ai-help message

[spark shard]

Wait i will try it again on another one

[low shard]

oh you're trying to clean audio? i thought you were talking about wokada for realtime voice changing?

[golden walrus]

[spark shard]

nope still doesnt work

[low shard]

yes, pls read https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window

[golden walrus]

i mean, it's pretty much the same

[low shard]

what browser are u using?

[golden walrus]

since some people said my output is a bit cracking

while my playback is clean

[spark shard]

I tried Opera and firefox

[simple ore]

so why are you downloading the thing for 5000 series then?

[pastel oak]

You probably use VB Cable instead of Line 1 (VAC) , it causes crackles

[golden walrus]

VB cable is better ?

[pastel oak]

no

[golden walrus]

.___. but i read somewhere

oh okay

[pastel oak]

vac download link is on the guide

[low shard]

Opera usually causes issues, have you tired chrome? also be sure u gave it microphone perms

[golden walrus]

no lah, i use VAC

[pastel oak]

Then do the crackle fix on the guide

[golden walrus]

let me check

[low shard]

ohh vonovox, i haven't tried it yet, seems fine tho logiically speaking

[golden walrus]

it supports spin

which is a mile better than cvec for my language

[spark shard]

Nope chrome doesnt work either and i gave it mic perms on all 3

[golden walrus]

wait, hold on, which one is crackle issue

[woeful adder]

its not working 😭 can you give me image perm?

[golden walrus]

i mean, this only occur to some people

i check it on my laptop and phone

it's all clean

[low shard]

!give-media-perms 1h @woeful adder

pls elaborate

[woeful adder]

what did i do wrong? i cant hear myself

[low shard]

vb audio cable is uninstalled right?

[pastel oak]

the audio setup is making me insanely depressed

[low shard]

uninstall vb audio cable

[pastel oak]

input: normal mic
output: cable
monitor (to hear urself if you want, optional): headphones

[low shard]

it causes issues

[low shard]

and she should use vac lite

not vb audio cable

[woeful adder]

oh what i download then

[spark shard]

jup

[low shard]

gpu: ur rtx 2060

extra: 2.7
chunk: 90ms

f0: rmvpe without onnx

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

always read the guide carefully, missing a step can cause issues

[woeful adder]

ookaaayy

[woeful adder]

i uninstall this thing right?

[low shard]

yes

[woeful adder]

theres many setups which one?

[low shard]

be sure also to not skip this, its very important

[spark shard]

@low shard i just tried restarting AND IT WORKS

Thanks so much for helping and sorry for being a pain in the ass the whole day

[woeful adder]

ahahhahaa

at first with my friend

2 years ago when i downloaded

i didnt know this

i was at terror

😭

i was like "WHY I CANT HEAR SHIT WTF DID I DOWNLOAD"

[low shard]

You're welcome and dw

[low shard]

yeah u gotta do things carefully cus AI isn't like installing whatsapp

[woeful adder]

it should be fine now right?

[low shard]

yes, but you still didn't uninstall vb audio cable

[woeful adder]

i did

it still shows

idk why

[low shard]

lmao what u been using this program for in 2 years

[low shard]

maybe try restarting

[woeful adder]

ill do it later its not that something that big

its not a virus right? so

no need to worries

[woeful adder]

i mean

i didnt use it

i just left it

[low shard]

its not, just be careful that it may cause issues in input/output, so thats why

so if u get any issues, uninstall it and reboot and retry using the program

[woeful adder]

uhm

its still not working

i did everything right tho i guess?

oh gpu

i saw that rn

[low shard]

nope

monitor: optional, headphones to hear urself
gpu: ur rtx 2060
f0 det: rmvpe without onnx
extra: 2.7
chunk: 90ms

[woeful adder]

it doesnt let me click rmvpe thing

oh i can now

OKAY

OKAY

STUPID ME HAHA

[low shard]

you need to click stop

before changing settings

[woeful adder]

okay okay i understood

i am stupid as hell so sorry

[low shard]

dw we are helpers to help people lmao

[woeful adder]

I CANT EVEN MAKE THIS CORRECTLY ITS SO HARD TO MAKE 90

[woeful adder]

yeah true true

[low shard]

it doesnt need to be exact

just around it

[woeful adder]

waahggg

anyways thanks its all working

saved it

[low shard]

on wokada, you can optionally:

• Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Reduce the delay via: https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

need any other help?

[woeful adder]

is this a guide or do ineed to download something?

[low shard]

if you're talking about the 3rd optional tip, it's a guide for less delay

[woeful adder]

ahh i see

too lazy for that pheeww

i dont need any other help thanks tho

do you get paid for this?

[low shard]

you're welcome!

[vague karma]

yo i need some help downloading

[simple ore]

-rt

[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

[vague karma]

need help i dont have amd or nvidia

[simple ore]

then don't bother

[vague karma]

can i use one even tho i dont have it \

[simple ore]

with a modern CPU, not something from 5+ years ago, you can run it on CPU, but nothing else.. no games

[vague karma]

so i cant use it or

[simple ore]

what do you have?

[vague karma]

uhh idk

can you send me a preferd version

[simple ore]

open

CPU name and GPU name

[vague karma]

i have amd

[simple ore]

details

you can run the voice changer on CPU, but you would not be able to play games or anything else at the same time

[vague karma]

why\

[low shard]

you want realtime voice changer for calls?

reminder, this is a general ai server, not just voice

[vague karma]

uhh just whatever need one for games and stuff can you send me a link to download it

[low shard]

that's bad, you sure you don't got any dedicated gpu?

[low shard]

if you dont got any other gpu, your pc is extremely way too weak

[vague karma]

@low shard which one could i run

[simple ore]

You ask "Can I buy an elephant?", I ask, "where do you live?", you say, "I live in the house", I ask "How big is your house?", you say, "very small", I say, "you can buy an elephant, but you would have to sleep in the corner on the floor.", you ask "why?"

https://github.com/deiteris/voice-changer/releases/download/b2332/voice-changer-windows-amd64-dml.zip

[low shard]

You got 3 options:

• Buy a better pc
• Run it locally (on ur pc) using the CPU mode of the wokada fork which has better performance https://rentry.co/ForkVoiceChangerGuide (but this isn't suggested as it could be unstable)
• Use **cloud **(remote good pc):

Cloud:

• Kaggles (30 hours weekly of better GPUs, T4x2 & P100, harder to use, requires an account and a phone number):
  • W-Okada's Deiteris' Fork Voice Changer Kaggle

You can't expect to run AI on ancient hardware

AI is an expensive task to run

Chatgpt runs on cloud, it uses a server with pcs that cost more than houses, all the sites with ai you see are just running on cloud, while the program runs on your hardware

[vague karma]

well which one do i dowload

the one the ai sent or

[simple ore]

I gave you the link

[low shard]

none, read what I said above

[vague karma]

dont get any

[low shard]

your pc is too weak

either buy a new pc or use cloud

[low shard]

his pc is too weak to run it with games on a cpu

it would be wasted time to download it

[vague karma]

can i run it without games

[simple ore]

5700g is not too bad

but as I said, wont be able to do anything else

no gaming

[vague karma]

i wont use it with games

[low shard]

you can try, but dont expect anything with a low delay

why don't you use cloud? you get limited time but it will run on a better pc, but hey your choice

[vague karma]

wait what is could

[crude flame]

😭

[low shard]

i mean cloud would be better if he doesn't use it more than 30 hours weekly

[vague karma]

uhh can you just send me cloud link

[low shard]

I already did #✨│ai-help message

plus it's a guide, you must read the guide to understand the program

[vague karma]

so read that guide

[low shard]

the cloud guide, yes

you're going to have 30 hours for free weekly of cloud

[vague karma]

is that the top link

[low shard]

and you need to use your phone number to verify your kaggle cloud account

[low shard]

no, the bottom one

read carefully, missing steps will just cause you more trouble

the program isn't user friendly, you must read else you will just get even more confused

it isn't a 1 click button program, it's AI

[vague karma]

its fine my pc dont work aint good enough

[low shard]

cloud runs on a remote good pc, it doesn't matter how good your pc

local runs on your hardware instead, which is why we suggesting cloud and not local

[wary shoal]

golly does any one know why i keep getting this error when i try to train a voice model on Applio? "The parameters of the pretrain model such as the sample rate or architecture do not match the selected model."

i got the sample rates matched i believe

[analog obsidian]

trying to use refinegan/mrf but using a hifigan pretrain instead

are u trying to use refinegan?

[wary shoal]

im using the hifigan one i believe

KLM- HFG right?

[analog obsidian]

wait

klm 4.9 hifigan?

[wary shoal]

its the KLM 4 that came with the application

[analog obsidian]

tell me the name of the pretrain pls

exact name

[wary shoal]

oh its the KLM HIfigan final version

48k

[analog obsidian]

are you sure you set 48000 in the first preprocessing step?

[wary shoal]

i believe so let me try it again

[analog obsidian]

[wary shoal]

dang it crashed again, its weird i got it to work last night

[analog obsidian]

[wary shoal]

now its giving me this message on the CMD prompt 😭 RuntimeError: The expanded size of the tensor (17280) must match the existing size (2640) at non-singleton dimension 1. Target sizes: [1, 17280]. Tensor sizes: [2640]

[analog obsidian]

mismatched sample rate

are you sure your audio is true 48k?

[wary shoal]

im gonna check the properties on each audio file and double check

[analog obsidian]

(this is not the real cause of the issue by the way, just asking because it's bad to train using the incorrect sample rate)

[analog obsidian]

no

thats not how you see the sample rate