✨│ai-help

[low shard]

The file oath shows 3.2.8, or did you just rename it

Colab is a cloud (remote good PC) service only for people with a bad pc

Could you please tell me your PC GPU first to see if it's good enough for local

[elder coral]

i didn't open my pc

[low shard]

Are you saying you don't have a PC and only a phone ?

[elder coral]

i'll do it tommorow

if i have a bad oc for colab what should i use instead

[low shard]

There are 2 ways to use AI

Local = runs in your PC but you need a good PC GPU

Cloud = remote good PC, but it has limited time on free tier, for example colab has only 4 hours max of GPU daily but it's random and can be even way less

[low shard]

Colab is an option for people with bad PC

If you have a good GPU you can do it locally

If you have a bad one, you will have to use colab

I'm asking because cloud services like colab are being very unstable those days and have limited times

We have seen people with a 2k PC still using cloud, which is kinda useless

And colab isn't even the only method for cloud btw

I'm asking for your PC GPU because if it's good enough you won't have to worry about the cloud limitations

[simple ore]

the reason is simple - there's torch >2.4.0 that does not allow torch.load without weights only parameter

there should not be 2.4+ installed

that may happen with non-rvc model that has no 'config' entry

[astral shell]

ye i renamed

[low shard]

Share the model download link

You might be using a non rvc model as noobies said

[astral shell]

Found it here https://huggingface.co/QuickWick/Music-AI-Voices/tree/main

[low shard]

Which

[astral shell]

took me a bit of scrolling, pop smoke

[low shard]

Those aren't for RVC

This repository is really old

Might be so vits SVC 4.0 or rvc v1

Did you find it in a YouTube tut or smt

[astral shell]

Ye legit

well damn thats a bummer

[low shard]

Everything you have seen on YouTube about RVC is completely outdated

You don't know how many awful settings and programs are there that are abandoned ASF lol

They are 2 years behind if not more

AI progresses at Sonic speed

Video tutorials can't be updated easily

[astral shell]

ye thats true

[low shard]

Which is why we do written guides

[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:

• #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

[pliant pier]

is there a way to have it in a separate application instead of it keep launching through opra?

[astral shell]

im planning to make my own voice model, id want it to be pretty accurate, any recomendation on how long the clips should be?

[low shard]

Elaborate

What program are you talking about? What guide are you following? What's your PC GPU? What's the issue?

Sorry if it seems rude but we can't help without information

This is a general AI support channel so we support more than a single program

[low shard]

Be sure they are extremely clean, at least 5-10 minutes long, wav as its lossy, and use the tensorboard

[pliant pier]

so for the old W-Okada it had its own "app" where it didnt use my browser but I switched to the newest version but I loads it like this

tf I cant uplead an image?

[astral shell]

How long should I train it for? How many epochs n stuff

[low shard]

Oh you mean original wokada and wooada deiteris fork

Both programs use a web user interface, the only difference is that original made its own "browser" window

Iirc it was removed in deiteris fork for better performance

So nope you can't make it its own window

Also if you're using operaGX, that browsers users reported issues with it for wokada, and it consumes more RAM for it's fancy effects

We suggest using chrome or firefox

[low shard]

!give-media-perms 1h @pliant pier

May u want me to check your settings meanwhile?

[pliant pier]

sure

I got it working just as good as I used too, still kinda monotone and robotic though

[low shard]

There isn't a right amount, epochs are just a unit of measurement of the training cycle
All you need to do is check how the training is going via the tensorboard, you can check also the docs https://docs.aihub.gg/

[low shard]

I would need to see all the settings, scroll

Also I would suggest you to use chrome or Firefox to not run into issues

[low shard]

Uncheck sup1
Check sup2 if ur having noise issues

Uninstall vb audio cable, it gives random issues reported by users (such as it just stopping to work for no reason lmao)
Please get vac lite from the 3rd step of the fork guide

Extra should be 2.7
Chunk 192, but this can change based on what program you're going to use wokada with, if you're having issues, just set the chunk to an higher value than the perf at the top left while running (a bit higher)

[pliant pier]

ah I don't use virtual audio cable

[low shard]

Its better you use vac lite rather than vb audio

Or is something else wrong ?

[pliant pier]

wait vac lite?

[pliant pier]

also do you mind sending me the guide?

[tame mica]

and how do i specify it to dl >2.4 next time

ts the nb i used https://www.kaggle.com/code/shiromiya/applio-public

[simple ore]

i can only suggest importing a notebook directly, you're using something very outdated

at least you need to get correct torch

!uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q

add it after this

[orchid flint]

I have a more general question, I have heard and know from personal experience that RVC has a hard time handling voices with a more robotic tone. However, what I don’t understand is why some robotic voices will train fine while others will struggle. I have had success with characters like FL4K from borderlands and Legion from mass effect but have had issues with Pathfinder from apex legends. Is it a dataset issue? I have about an hour of data for legion and pathfinder yet legion trains fine while pathfinder starts to OT at around 100 epochs. They don’t have any echo or reverb just vocal processing, so I’m not sure why I keep keeping these unpredictable results. There has to be something that I’m doing wrong. Anyone have any ideas?

[simple ore]

when you have a pretrain with 50+ hours on normal people voices speaking, and then you try to train a model with a tiny dataset that is not quite like that, the model needs more time to readjust

dont think it it would overtrain at just 100 epochs, but it wont perform as good as a model with a normal speaking voice

[low shard]

Yes

[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

I thought u had it already since u got the program

[orchid flint]

Yeah, that makes sense. but I still want to look into this more. Do you have any pre-trains that you could suggest? I want to see if I can find one that would work better than OG for my purposes. Especially since I still can’t seem to pinpoint what factors RVC/pretrains specifically don’t like.

[simple ore]

well, any synthetic distortions wont be good

[orchid flint]

Yeah, I have seen that it can work though.

[raven perch]

It is personal.

[meager axle]

Hi
I have a proplem with " Voice Ai "
It tells me: "GET THE ULTIMATE AI VOICE EXPERIENCE"
Whats the deal with it? Isn't he for free??

[knotty moth]

it is freemium and utter garbage

[meager axle]

So why i cant use it?

I cant take this page off
There is no " X " sign in it
I even tried by hitting the ESC but didnt work

[low shard]

it's paid and also stays in the background using ur pc resources

don't follow yt tuts for realtime voice changing lol

what's your pc gpu?

[hallow thistle]

So what? Here are issues about Voice.ai:

1. Voice.ai can use your PC resources for their cryptocurrencies
2. It works worse than W-Okada

[meager axle]

I core 5 13th

[hallow thistle]

That's PC CPU, not GPU.

[low shard]

that's a cpu not gpu

those are 2 different things

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

[meager axle]

Oh sorry , wait i'll check it

[hallow thistle]

Read again, GPU = graphics processing unit, and CPU = central processing unit; these two are different things. To check your PC GPU, open Task Manager, go to the performance tab, spot where GPU 0 or 1 is in the left panel, and click on it to show its full name on the right panel.

[meager axle]

Intel(R) UHD Graphucs

[low shard]

that's really bad

it's integrated graphics, which is weak

are there any other gpus?

[hallow thistle]

Any other than this one or just that?

[low shard]

!give-media-perms 1h @meager axle

@meager axle u can send a screenshot too if you want

[meager axle]

No

Its a laptob

[hallow thistle]

If it's just Intel UHD Graphics, that doesn't sound good.

There's another option, use a cloud service like Kaggle instead.

[low shard]

your pc is too weak to run it locally (runs on ur hardware), you will need a cloud (remote good pc) method which has limited free time

[meager axle]

So isnt there any Ai voice changer i can use?

[low shard]

most cloud methods are broken right now

the only working one is Wokada deiteris fork Kaggle

[hallow thistle]

There is.

[low shard]

locally no, that pc is really weak, on cloud yes

[hallow thistle]

Don't worry about it. You can still use W-Okada, it just doesn't need to run locally on your PC for this time. You'll have to register with your phone number to use Kaggle.

[low shard]

if you want an explaination:

RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models

Wokada = uses RVC for realtime inference

original wokada is made by wok
wokada deiteris fork is made by deiteris
deiteris is the most suggested since it has more improvements

locally = it runs on your pc hardware, like opening a game

cloud = using a remote good pc hardware, like chatgpt

Wokada and RVC are both Open Source, meaning the code is free and public for everyone

laptops aren't that powerful, and your hardware isn't enough to do it locally, so you need to use cloud

[meager axle]

Thanks for helping me

You r very kind

[low shard]

you're welcome and lmk for any further issues

[knotty moth]

it wants to take your money

[low shard]

along with the link I gave you, you need to also follow only this step of this guide to get a VAC installed in your pc https://rentry.co/forkvoicechangerguide#virtual-audio-cable

you need to follow only that part, not the rest

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

[meager axle]

Ok i will rty

Try *

[austere fern]

can someone help me wiht something i need?

[hallow thistle]

Also the Voice.ai ain't even free.

[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.

[low shard]

elaborate

[hallow thistle]

You're looking for W-Okada. What is your PC GPU? And which W-Okada version are you using?

[low shard]

@austere fern #🧬│ai-chat message I saw you said both rvc and wokada

those are 2 different programs

RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models

Wokada = uses RVC for realtime inference

please be sure to mention your pc gpu, what program you want, what you want to do and what's the issue

also be sure to never use video tuts for rvc/wokada

[austere fern]

what do you mean never use video tuts

[hallow thistle]

"The audio isn't working" always happens to those who followed tutorial videos on YouTube as their main guide. Certain tutorial video uploaders really put Discord invite link to here in their description as an excuse.

[austere fern]

i did use a tut from yt

[hallow thistle]

Most of the videos there only telling you to download and install the "original" version of W-Okada, which is so far outdated in the end.

[austere fern]

for me it says 2.0.76 beta cuda

[hallow thistle]

Now what's up with your PC GPU?

[low shard]

they are all outdated

you should always check the date of a video before using it

ai progresses at sonic speed

in fact you got original wokada

instead of the suggested wokada deiteris fork

[austere fern]

i have nvidia geforce rtx 2070

[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]

read up the 1st link

the wokada deiteris fork

you can uninstall everything you got off youtube tuts

if u installed vb audio cable, that gives random issues on windows as reported by users

[austere fern]

oh so i unistall everything

[hallow thistle]

Download and use the better W-Okada as Nick sent to you instead. https://rentry.co/ForkVoiceChangerGuide#download-nvidia-on-windows

[low shard]

yes, everything you got off youtube tut

read the deiteris fork guide

[austere fern]

okay

i did it

unistalled everythimg

[low shard]

alr, u can read the guide now

[austere fern]

i cant understand

[low shard]

what can't you understand ?

[austere fern]

the guide and how to download the latest

[hallow thistle]

You can't understand? That sounds like a skill issue.

Don't expect me to tell you everything about W-Okada. I can only give you some ideas and you must think what you gonna do next.

[low shard]

you got an nvidia gpu and are probably on windows, follow that parts of the guide

[hallow thistle]

To download W-Okada, head over this "Download NVIDIA on Windows". There's a hyperlink text that says "nvidia-b2332 (click here to download)", click on it.

Ignore the one that says "NVIDIA RTX 5000" below. That one is for NVIDIA GeForce RTX 50 series GPU. You have NVIDIA GeForce RTX 2070, so no need to download that one.

If you see this, click on the highlighted "download" word to download W-Okada zip.

[brittle wing]

wtf is this

[simple ore]

ancient version

download the compiled package

[brittle wing]

fuck

wheres that

[simple ore]

https://huggingface.co/IAHispano/Applio/tree/main/Compiled/Windows

[brittle wing]

thank you bro

[simple ore]

and unzip it to just C:\Applio

not a 200+ character long path

[brittle wing]

so if its on my documents folder its wrong?

[brittle wing]

C:\Applio\ApplioV3.2.8-bugfix\env

so this is good now right?

[simple ore]

well, one applio too many again

[brittle wing]

alr

[limber pier]

hi

Why does this sound keep coming up? "TRY OUL"

[simple ore]

use vac lite

[limber pier]

what is it

[simple ore]

you're using trial version so it periodically adds trial audio

[low shard]

you basically downloaded vac trial instead of vac lite

[limber pier]

thank u i kissing youuuuuuuus

[jade orchid]

how do i get pip

[finite parcel]

So they are better than all the AIs i mentioned?

[low shard]

Are you following a YouTube tutorial for Rvc or wokada

Those are old asf

Don't follow them

U don't need to install python anymore for them

Tell our PC GPU and what u want to do

[low shard]

That's my opinion yeah

[jade orchid]

yes i am

what tips could i follow

for it to work

NVIDIA GeForce RTX 3070 Ti this is my gpu

and i want to use rvc

but evrytime i click start conversion or sum it just freezes and crashes

[simple ore]

pip comes with python, but depending on what you're actually using you dont need to use a standalone pip

[jade orchid]

thank you

[brisk steeple]

hey!! i have a 600 page book explaining a course creation / learning method and I want to use LLM to make courses using that methodology, which model should i use and whats the best way to write the inputs and polish the outputs

[red hare]

training an czech voice, got studio clean vocals, 1h 20min data, used snowiev3, 100epochs what yall think? dry vs wet here,
should i continue training more epochs?
https://krakenfiles.com/view/OgQ9Ud4BX6/file.html - wet
https://krakenfiles.com/view/6CMCjbDGlu/file.html - dry

[simple ore]

snowie is a weird choice, especially if you're going with a singing set

examples sound autotuned

[red hare]

used snowie cuz of the accent, czech and russian is kinda close, and its a rap vocal, also tried titan medium

[simple ore]

if you dont use an index, the accent will be minimal with any pretrain

[red hare]

okay, so what recommandations would you give me?, should i just train without pretrain? idk what pretrain to use so i used the snowie one, im new to this soo thats that haha

[simple ore]

try klm 4.9

https://discord.com/channels/1159260121998827560/1339155300720054316

[red hare]

sure thanks :))

[low shard]

cooked

[low shard]

RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models

Wokada = uses RVC for realtime inference

you want rvc right?

[low shard]

Nice

[jolly tangle]

Stupid question how do i use this when training?

[vast elm]

so basically you download the G and D pth pretrain files, you put them in the folder assets/pretrained_v2, now when training the model there should be a path input for the pretrained model you want to use at the bottom left of the page

[jolly tangle]

Thx

[simple ore]

pretraineds_custom folder

not v2

[analog obsidian]

assets/pretrained_v2 is the location of the original pretrains in mainline

in applio the structure is:

rvc/models/pretraineds/custom

[red hare]

what batch size should i do if i have about 1hour and 20 minutes of data?

[analog obsidian]

8

[knotty moth]

read this guide https://discord.com/channels/1159260121998827560/1262390831998963713

[vast elm]

Non Applio/Other RVCs Users :
 - Download the pretrain
 - Be sure to use the right sample target rate based on the pretrain you are using and if it’s also the one of your dataset, and that they are also the same language.
- Put them into the pretrained_v2 folder in your RVC.
 - Open RVC, Be sure to put version v2 and set the G & D pth of the pretrain you downloaded and want to use.

Oh well by default i thought he wasnt using apllio, but since a large majority of the people here use Applio i should've thought of mentioning that too, thanks for the correction, i appreciate you pointing that out

[simple ore]

train

[elder coral]

bro what should i use instead of colab to make ai models

my model maker disappeared when i left

i joined this server since 2023

[trim sparrow]

is this normal ? ```
warnings.warn("Detected call of lr_scheduler.step() before optimizer.step(). "
INFO:SingerJoshing:Train Epoch: 2 [0%]
INFO:SingerJoshing:[21, 9.99875e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=6.169
INFO:SingerJoshing:====> Epoch: 2 [2025-04-15 21:54:15] | (0:00:48.283193)
INFO:SingerJoshing:Train Epoch: 3 [0%]
INFO:SingerJoshing:[42, 9.99750015625e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=6.866
INFO:SingerJoshing:====> Epoch: 3 [2025-04-15 21:55:02] | (0:00:47.281693)
INFO:SingerJoshing:Train Epoch: 4 [0%]
INFO:SingerJoshing:[63, 9.996250468730469e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=8.847
INFO:SingerJoshing:====> Epoch: 4 [2025-04-15 21:55:50] | (0:00:48.020638)
INFO:SingerJoshing:Train Epoch: 5 [0%]
INFO:SingerJoshing:[84, 9.995000937421877e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=7.718
INFO:SingerJoshing:Saving model and optimizer state at epoch 5 to ./logs\SingerJoshing\G_2333333.pth
INFO:SingerJoshing:Saving model and optimizer state at epoch 5 to ./logs\SingerJoshing\D_2333333.pth
INFO:SingerJoshing:saving ckpt SingerJoshing_e5:Success.
INFO:SingerJoshing:====> Epoch: 5 [2025-04-15 21:56:40] | (0:00:49.878488)
INFO:SingerJoshing:Train Epoch: 6 [0%]
INFO:SingerJoshing:[105, 9.993751562304699e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=7.401
INFO:SingerJoshing:====> Epoch: 6 [2025-04-15 21:57:28] | (0:00:47.433959)
INFO:SingerJoshing:Train Epoch: 7 [0%]
INFO:SingerJoshing:[126, 9.99250234335941e-05]
INFO:SingerJoshing:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=7.861

[steel forge]

Applio

https://github.com/IAHispano/Applio/releases

[elder coral]

but i'm gonna check my gpu first

[steel forge]

Its the same thing as what you used before prob, just more modern

[steel forge]

Ye

[elder coral]

whether if it is good or not?

[steel forge]

My 1060 trained models ok

[steel forge]

What GPU?

[elder coral]

My gpu is nvidia geforce RTX 3050

I'm already here since 2023

but i forgot all of these techniques

[steel forge]

Oh yea

You're good

You'll be making models pretty quick

(Hours)

[elder coral]

now what command should i use to run applio

[steel forge]

Its the run-applio.bat in the folder you downloaded

Read the instructions on the download page

[elder coral]

i have to install first

[steel forge]

I don't think so

[elder coral]

my model maker role has been removed

[steel forge]

If you downloaded the compiled zip, it should just be run-applio.bat

[elder coral]

yes

i know

look at this error after installing all resources

wtf is this we can't wait that long

[candid lichen]

You guys got a photo upscaler, tryna make my ai photos a bit more high quality

[elder coral]

how to fix it

[sharp glacier]

I'm to stupid to remember the settings that are correct. Can someone refresh my memory on all correct settings.

[pastel oak]

Input mic
Output virtual cable

Then in discord and games:
Input virtual cable
Output headphones

[pastel oak]

If youre using stable diffusion you can use hi-res in the options

[pastel oak]

Yes your model is training rn

[trim sparrow]

cor some reason when i tried using it i would just get error

[sharp glacier]

okay what do I change to what.

first= Mic
2nd=
3rd=

----> = ?

[pastel oak]

I said 2nd is virtual cable already

3rd is playback to your headphones so that doesnt matter, if u want to hear yourself then use headphones/speakers

[candid lichen]

Iv heard of stable diffusion I just don't know if my laptop can handle it, I can look up the specs and see

[sharp glacier]

sorry i'm just slow and don't use it much and havn't in a long while. so the last one can be ethier or?

[pastel oak]

Yes, its optional

Leave it at none if you want

[sharp glacier]

this is what I have so far

[candid lichen]

Damn even 8gb ain't enough apparently that's tough

[pastel oak]

What kinda error? You mean the red box "Error" when you do inference? Show the error code in the cmd

[pastel oak]

Whats ur gpu name

[knotty moth]

SD 1.5 is pretty lightweight for 4 GB cards, and XL can run on 6 GB

[trim sparrow]

i got ya let me finish eating n ill hop back on my computer it was related to ffmpef i think

[candid lichen]

I'll see rq

[pastel oak]

🤔 doesnt make sense to have this error happening now , usually the program doesnt even start up if ffmpeg isnt installed. Which rvc program do you have

^ and for referenfe ffmpeg is installed automatically if you downloaded precompiled rvc (meaning you didnt manually set up an environment to run rvc)

[candid lichen]

Intel UHD Graphics 605

[knotty moth]

unusable, so it will run in cpu mode

[sharp glacier]

this is what i was going off of but it no longer works.

can someone just send me a visual aid please.. I'm not good with setting things up

[knotty moth]

that issue occurs on mangio under certain GTX/RTX 20-series gpus, and the results will be nothing different from the pretrain instead of your dataset

go get the better one https://docs.aihub.gg/rvc/local/applio/

[candid lichen]

Will that other program work fine then?

Oh is that for stable diffusion? I have no idea what I'm doing I won't lie here

[sharp glacier]

okay let me try this slower so I can understand how the new version works.
this I set to which of these options?

[pastel oak]

Input: microphone (probably what you highlighted yes)
Output: Line 1
Monitor: none

And on discord and games
Input: Line 1
Output: headphones/speakers

Should use a headset since playing back on Speakers picks up all your pc audio back into your mic, so it echoes fyi

[pastel oak]

Stable diffusion wont run fine. What program are you using atm?

Any ai image creation wont run locally with intel uhd

[candid lichen]

Iv been doing ai image creation all from my phone on a website

I was trying to find different either spaces on hugging face or programs to use a different model but kinda limited by having a budget laptop, even my phones specs are better then it lol

[steel forge]

I'm a noob. All I can suggest is weight.gg's image generation

I still use automatic1111 SD on my PC lul. Pretty sure that's old

[candid lichen]

I did for a while then I by chance found a website you can use multiple different models to generate images, I used the weights gg app for a while I just didn't like the character limit is literally all, but the results it made were really well

And it does it all at the same time

[knotty moth]

I'd suggest searching some SD/comfyui colab/kaggle notebook

it still runs much faster than cpu mode for SD 1.5

[sharp glacier]

thanks

[candid lichen]

Oh a colab thing? I think I attempted to try that before but I might have done it wrong, I took code from hugging face and put in collab, to no avail it didn't work, literally no idea on how this works

I think also you needed a license to use curtain models if I understand it correctly

[knotty moth]

it is kinda tricky to set it up since you've to avoid getting banned

[candid lichen]

Wait what?... Why would you get banned?

[knotty moth]

the webui thing, so it'd be some obfuscation trick and ngrok tunnel setup

[candid lichen]

Sounds like alot more work then I wanna do just to upscale lol, but anyways thank you for tryna help I appreciate it

Knowing me id probably get banned just by even trying to figure out how to get around that if I'm being fr

[orchid zealot]

guys

suggest me any open source HeyGen alternaative

[low shard]

https://github.com/R3gm/SoniTranslate

[barren thunder]

Hi, can someone help me wih this error on mainline collab plz:
Traceback (most recent call last):
File "/content/training/runmain.py", line 3, in <module>
from dotenv import load_dotenv
ModuleNotFoundError: No module named 'dotenv'

[orchid zealot]

pip install python-dotenv

[barren thunder]

yes bu it's already satisfied

[orchid zealot]

pip install --force-reinstall python-dotenv

[knotty moth]

https://tenor.com/view/bug-fix-fixing-bugs-in-your-code-bugs-code-sinking-gif-17779185

[knotty moth]

the colab notebook is dead and will never be fixed

[odd isle]

where can i download the latest version of w-okada?

the version i have right now is outdated

[knotty moth]

ftp?

[low shard]

What's your PC GPU and what do you want to do

[eternal plinth]

is there zero two voice?

[azure grail]

Does anyone have a realistic Bad Bunny model?

[pastel oak]

@eternal plinth @azure grail

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/

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/

[hallow thistle]

Find one in #1175430844685484042, search by keyword, click on any of these, and then hear the sample audios there.

[hallow thistle]

I think you used the "original" version of W-Okada earlier, which is indeed outdated in the end. What is your PC GPU? Because W-Okada works best with a decent GPU.

[odd isle]

3060 ti and why?

[odd isle]

3060 ti

[low shard]

Nice

-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, read Wokada deiteris fork

@odd isle about the why, it's because the program is open source and runs locally on your hardware

It's not CHATGPT where you use a site with a remote good PC cloud

You can't run ai locally on a bad pc

It's like games, you have to check the requirements before, and this is for basically any pc programs, like video editing

[odd isle]

thanks

[low shard]

Yw and lmk

[low shard]

@lofty lichen about the message you sent, I also want to remind you that there isn't any working paper space RVC notebook at the moment

[lofty lichen]

applio

[hallow thistle]

Applio is RVC program.

[low shard]

Also you can check https://www.paperspace.com/pricing

[lofty lichen]

https://docs.applio.org/applio/getting-started/other-alternatives

at the bottom

[low shard]

@simple ore did any applio staff check if applio works correctly on paper space still?

Last time any staffer here tried paper space it didn't work well with just 3 commands and there were issues in the environments, which is why we also don't have paper space guides for RVC mainline nor Applio, along the reason that most staff doesn't use paper space anymore

I could be wrong but just checking

[raven perch]

Hey GM.

[simple ore]

no idea, as long as you can run usual linux stuff it should be doable

[lofty lichen]

If the pretrain im using doesnt support my language it wont work? im talking but it says random stuff lol

[hallow thistle]

What is the Paperspace even for? Because most of the time I use Google Colab and Kaggle, and also run locally.

[hallow thistle]

RVC will work with any language, unless you mean either index file that stores accent of voice model or some kind of TTS model.

[knotty moth]

depend on the accent and speaking style, sometimes it may sound too far away

[sacred geyser]

ok i have been trying this a 9 different ai picture generators they all output the same thing lol , i want a photo of a video card on a piece of paper on desk no matter how it word it it ends up being an actual video card on top of a piece of paper on a desk sample of prompt:
create photorealistic image of a piece of white paper lying on a wooden desk. On the paper is a high-quality, detailed printed image of an NVIDIA RTX 5060 video card. The scene is well-lit with natural light, and the focus is sharp, showing both the texture of the paper and the desk. The photo is taken from a slightly angled, overhead perspective.

[raven perch]

Does this server have the option to set-up or create support tickets?

In case you need troubleshooting but you don’t want 500 peeping Tom’s’/ lurkers?

[limpid trout]

Alright that's it,
I want Voice changer which isn't like Real time voice changer For the content I require not real time voice changer because it won't run well in CPU based PC (I have laptop) so I want a program or project that only does voice changing no real time, just drag the folder in and wait till it does the thing .

[limpid trout]

And no, I don't want websites which does the same. I want offline based in my laptop, voice conversation

More like local

[slender osprey]

My microphone stops working when I put it as requested in the manual, is there no way to leave it with cd quality and change the application?

[vale osprey]

-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

[slender osprey]

16 GB ram, Intel Core i5-8400 - CPU @ 2.80GHz - AMD Radeo n RX 580 2048SP - Storage 447 - 500 MB Internet - I can't find the ideal RVC voice ai for my computer, could anyone help me?

[gray cobalt]

I’m not sure why, but I’m pretty sure my okada is only running on CPU even tho I set it to GPU

[simple ore]

realtime?

[low shard]

Hey

[low shard]

could you please ask for an applio staff that has paper space to try it out? last time it someone tested it they also said common issues https://docs.google.com/document/d/1ooG2hJrfNNLUln0reTKKIOpNBjp53G0joak50H_sQhE/edit?tab=t.0

but this is outdated too 😭

[low shard]

yes via dming @vital hedge but usually tickets are meant for moderation things

[low shard]

what's your pc specs?

RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models

Wokada = uses RVC for realtime inference

I'm guessing you want RVC

[low shard]

elaborate:

• what's your pc gpu
• what do u want to do
• what tutorial link did u follow
• a screenshot of ur wokada

never use youtube tutorials for RVC & Wokada, they are old

[low shard]

RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models

Wokada = uses RVC for realtime inference

I'm guesssing you're talking about wokada
Elaborate the issue and send a screenshot of your wokada

[limpid trout]

Something you wouldn't believe.
But anyway retrieval based voice conversation maybe ok
How can I install it

[low shard]

i mean i asked to see if it can even run RVC

if it's a 10 year old laptop with 1gb of ram, don't expect it to run locally

it would be better you tell me so i can tell you if it can run and you don't waste time

[limpid trout]

4 GB ram i5 intel and cpu

I think

[low shard]

you can also search task manager on windows and check the performance tab showing a screenshot so we are sure that's your specs

just saying in case u aren't sure about your specs

[simple ore]

i aint paying for this

[limpid trout]

Hmmm, it's actually 8GB (7.88 usable) intel core i5 5300U CPU 2.30 GHz 64bit

[low shard]

ye me neither I was just asking if there's an applio staff that checks if it still works lol

[low shard]

it can run on your pc locally but it really won't be a good experience, meaning:

• slow
• can't do too long audios unless you split in some cases
• ofcourse, you can only inference (use the model). you could theoretically train (make models) but it could take a month (24/7 on) if not more for a single one

are you 100% sure you want to do it locally?

[limpid trout]

Yes

And besides I am getting new pc later. But now I wanna know how to install that retrieval based voice conversation

[low shard]

because you can do it for free on cloud (remote good pc) sites, like you don't have to pay since you could use kaggle that offers 30 hours of better gpu weekly for free

but it's your choice

[low shard]

You can use RVC locally, you can choose between:

• Applio: A fork (modified version) of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
• Mainline: The original RVC

[limpid trout]

I could, but
For some reason it doesn't work well. Especially with my network

[low shard]

I would personally suggest Applio

[low shard]

you don't need a 10ms ping network like cloud gaming

[simple ore]

free instance is not available and aint nobody gonna pay for this

[low shard]

like, you don't need a good network for it

[slender osprey]

Yes

[low shard]

the speed depends on the remote good pc, not your wifi

if you had issues, it's probably because you followed some old tutorial

[limpid trout]

I dunno.

[low shard]

@limpid trout
if you are ever interested to try cloud:

Train (make) RVC Models on cloud:

1. Prepare the Dataset
2. Setup RVC:
   Choose a cloud way to use RVC,

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio by Vidal (UI)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • Applio (UI)
  • Mainline (UI, No guide as of right now)

3. Be sure to know about the tensorboard

Google Colab = Easier but risk of getting disconnected
Kaggle = Harder but way more gpu time
If you are looking for the easiest way and for free, is using https://weights.com/ which ofc uses RVC

RVC Inference (use models) on pre-recorded audio on Cloud

You can use either:

• Weights.com: Easiest Possible Ever Automatic
• Ilaria RVC Zero: Fastest free on cloud
• Applio UI Colab: RVC Fork with some extra features like TTS
• RVC AI Cover Maker UI: Automatically Separates the Vocals and Instrumentals, converts the voice and mixes them back

NOTE: Mainline is currenty having issues on colab

[low shard]

I would personally suggest you to try cloud before going with local

I can help you out for any issues ofcourse

[limpid trout]

Hmm... Ok let's see

[low shard]

send a screenshot of wokada

!give-media-perms 1h @slender osprey

[low shard]

why is it on the site then

ofc im not trying to be rude, i'm just saying in case someone actually buys a paperspace sub (like that user was saying) and then finds out it doesn't work

[low shard]

outdated asf

that version is more than a year old

lemme guess, you followed a video tutorial?

because they are all outdated

only updated guides are the written ones

you can delete everything you got off that video tutorial

[slender osprey]

yes

[low shard]

along with vb audio cable since it gives random issues on windows

[low shard]

forget everything they said bc the settings are really messed up along with the program version

Wokada has 2 main versions:

• Original made by Wok
• Deiteris fork (modified version) made by Deiteris

each version has it's own updates

the latest deiteris fork has way better performance and quality than the 1 year old original wokada version you're using, especially since you're on AMD gpu

@slender osprey also, your rx 580 is going to be enough, but will have issues if you're planning to use it for games like marvel rivals

so mostly will work for vc and really low intensive games with lowest settings

-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]

@slender osprey read the 1st guide, wokada deiteris fork, and let me know how it goes

[slender osprey]

thanks

[low shard]

you're welcome

[simple ore]

does the method shown here https://docs.applio.org/applio/getting-started/other-alternatives not work?

[low shard]

oh that's different than the one shown before, the issue is that no one can check 😭

sorry if i seemed rude, I was just trying to suggest that maybe it shouldn't be in the site if no one can test nor know if it works now, i think you have seen how much you gotta update the colab and paper space might need updates too

anyways, it's not ai hub site, just my own curiosity and suggestion as I was afraid that user was going to complain (and waste money) if it didn't work and no staff could help since no staff has paper space

idek why paper space puts a free option if it hasn't been available since like a year, i checked it a few times and all the times it was never available lol

[simple ore]

as long as you can run normal python ai stuff, applio should not be a problem

but I cant verify

[slender osprey]

Just to know if I'm in the right place... is here?

[low shard]

nope, did you read fully the guide?

you should check https://rentry.co/forkvoicechangerguide#download-amd-intel-and-cpu-on-windows

please don't skip steps

be sure you also uninstalled your old wokada, vb audio cable and checked the 3rd step of the guide too

[slender osprey]

Okay, I did it

[low shard]

no you're downloading the wrong version

you need to download the amd/intel/cpu one

i sent you the link

nvidia version is made for nvidia gpus

[slender osprey]

Aaa

Sorry 😦

[low shard]

It's fine

@slender osprey how's it going?

[slender osprey]

I didn't understand what I did wrong

[low shard]

Dont use operagx

Many users reported it giving issues

Try chrome or firefox

Also uncheck sup1

[simple ore]

illegal combination of device is the use of different drivers (WDM mic, MME out)

you need to use the same type

[slender osprey]

k

ok*

[restive birch]

I have a question. When I'm using a game with voice chat. Did I have to put 'Line 1' as my microphone and my micro as the output in the game?

[slender osprey]

Now it's gone, is there a way to make it faster? Like a delay of 1 or 2 seconds?

[simple ore]

just change it to 3.11

[hollow rapids]

how to dowloand voice model?

[lofty lichen]

on ur website?

for copy

[simple ore]

export LD_LIBRARY_PATH=/usr/local/lib/python3.11/site-packages/nvidia/nvjitlink/lib:$LD_LIBRARY_PATH

[pastel oak]

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/

[pastel oak]

rx 580 is pretty bad, but theres a few options you can try out:

• use audio mode: SERVER
  S.R. 48000
  [Windows WASAPI] on every prefix, and then choose the same stuff as always. make sure all of them are windows wasapi, this cuts down some more delay
• change your chunk to as close as possible to the perf but still have it higher. PLEASE NOTE: amd gpus need to change from gpu to cpu back to gpu whenever you change chunk or extra to "reset back to full power". so if you change chunk, the perf will INCREASE which is "worse". so always do the gpu to cpu to gpu fix after changing chunk
• advanced settings: crossfade length 0.05 is technically faster than 0.1 but quality can get a little worse. i suggest trying it out

[slender osprey]

Like this?

[pastel oak]

Now after changing to gpu, what perf value do you get on these settings?

[dusty mortar]

while training a model how important is the batchsize really? i could use 32 which is far faster than 4gb or is the quality hit significant?

[lofty lichen]

does nothing

[slender osprey]

Thank you, the results were more satisfactory

[lofty lichen]

tried before make run-applio and after

[simple ore]

run the make command

this one sets the path to libraries

[lofty lichen]

make export LD_LIBRARY_PATH=/usr/local/lib/python3.11/site-packages/nvidia/nvjitlink/lib:$LD_LIBRARY_PATH

[simple ore]

no

make run-applio

[lofty lichen]

same error after doing it

and running make run-applio

[simple ore]

could you paste a text of the error message?

[lofty lichen]

root@nwb6j346kb:/notebooks/Applio# make run-applio
python app.py --share
Traceback (most recent call last):
File "/notebooks/Applio/app.py", line 21, in <module>
import rvc.lib.zluda
File "/notebooks/Applio/rvc/lib/zluda.py", line 1, in <module>
import torch
File "/usr/local/lib/python3.11/dist-packages/torch/init.py", line 239, in <module>
from torch._C import * # noqa: F403
^^^^^^^^^^^^^^^^^^^^^^
ImportError: /usr/local/lib/python3.11/dist-packages/torch/lib/../../nvidia/cusparse/lib/libcusparse.so.12: undefined symbol: __nvJitLinkAddData_12_1, version libnvJitLink.so.12
make: *** [Makefile:18: run-applio] Error 1

[simple ore]

run find /usr/ -name libnvJit*

[lofty lichen]

root@nwb6j346kb:/notebooks/Applio# find /usr/ -name libnvJit*
/usr/local/lib/python3.11/dist-packages/nvidia/nvjitlink/lib/libnvJitLink.so.12
/usr/local/cuda-12.0/targets/x86_64-linux/lib/libnvJitLink.so.12.0.76
/usr/local/cuda-12.0/targets/x86_64-linux/lib/libnvJitLink_static.a
/usr/local/cuda-12.0/targets/x86_64-linux/lib/libnvJitLink.so.12
/usr/local/cuda-12.0/targets/x86_64-linux/lib/libnvJitLink.so

[simple ore]

okay, export LD_LIBRARY_PATH=/usr/local/lib/python3.11/dist-packages/nvidia/nvjitlink/lib:$LD_LIBRARY_PATH, then make run-applio

sorry

no make for export

just export ...

[lofty lichen]

works

[simple ore]

@low shard so it seems to work with this fix

[low shard]

that's great to hear, I hope the fix gets updated in the site too

[simple ore]

done

[low shard]

great job

[frosty maple]

How to fix thi error
[VCClient] Access http://127.0.0.1:18888/
[VCClient] wait web server... http://127.0.0.1:18888/ ECONNRESET

[stark reef]

Guys I get errors every time I put in an inferencing voice 😭

I tried looking everywhere to fix it 😭😭

I’m on RVC WebUI

I even used their original models

And it’s still not working

It’ll process in the auto detect index path section and in the voiceless consonants section for like 2 seconds and then cover them in big red errors

And if I press convert I just get two more red errors

I’ve been at this for like 3 hours 🥲🥲🥲

[versed edge]

I have this issue with w-okada where the audio sounds fine on my monitor but in the game (In this case it's roblox), if there is loud in game audio playing, the voice changer would sound very muffled and bugged out. (Normal mic works fine regardless of in game audio so it has something to do with either the voice changer or the virtual audio cable)

[knotty moth]

not showing error messages in the console window means no help

and tell me if u want to show the screenshot

[stark reef]

I’m a lil confused idk what this means

I can’t send images here but I could dm it to u?

[knotty moth]

!give-media-perms 1h @stark reef

no response means ur cooked 😔

[stark reef]

AAA

okay one sec

[stark reef]

😭

I could record what it does exactly actually

[knotty moth]

wrong, not this shit but another one which is console window

it has black background with text only

[stark reef]

Gotcha

[knotty moth]

that kind of error message means unsupported gpu (perhaps older than GTX 10-series)

[stark reef]

OHHHHHHH

Okay so Idk anything abt gpus but from what I understand mines super old and outdated- so do I have to get a newer one for this to function essentially?

[knotty moth]

GTX 1050 Ti (or 4+ gb) is bare minimum possible, but RTX 2060 is recommended spec

[stark reef]

Thank you! <3

[elder coral]

@steel forge pls how to

basic

i fcking have nitro in june 2024

@lofty lichen how do you fix this glitch in applio

i download all resources from run.install.bat

[knotty moth]

try downloading the precompiled zip

[elder coral]

where

i have tried applio months ago and did not give me an error

how to download the precompiled zip

i lost my model maker when i left ai hub

[knotty moth]

ask google bro

[elder coral]

i already have a bugfix file but it still gave me an error

[knotty moth]

also opening run_install.bat when it previously work could break it

[elder coral]

i actually got these red errors and my gpu is good

its nvidia geforce rtx 3050

[hallow thistle]

Make sure you install Applio in the right directory like in D:/ or C:/Users/"your username"/Downloads/. Installing anything directly in desktop folder can cause several issues.

[elder coral]

ok

[simple ore]

just C:\Applio

not some crazy long path

[elder coral]

it got me an error in the downloads agaim

it protects the malware

[simple ore]

using some terrible antivirus?

[elder coral]

i turned off realtime protection in windows security

now my folder is in c:/

[hallow thistle]

If one of your antivirus programs detected Applio, it's a false positive. Unless you've download either of these nowhere.

[elder coral]

I've lost my model maker role since when i left the server

I have to make a model again

my version is outdated it didn't work

[potent bone]

my gpu is 3060 ti should i download that one?

[hallow thistle]

Yes.

[half meadow]

Gents, made model how upload

explain it to me like I'm a mentally handicapped caveman please and thank you

[knotty moth]

the easier way is to upload to weights.com

otherwise you need to get #outdated-model-maker-role and read its guidelines in order to publish in #1175430844685484042

[half meadow]

appreciate you pookie

upload complete, yippie

[elder coral]

After i installed applio why do i get this error at run-applio.bat why do i see a bunch of errors

My GPU is Nvidia RTX 3050 Ti

[knotty moth]

I thought you're using an 8K monitor but can't even read it with any upscaler

[elder coral]

yeah i reintalled again

there are no red errors

if i have a good gpu why do i see this error

[knotty moth]

try manual install from the repo's main branch https://github.com/IAHispano/Applio/tree/main

[elder coral]

I already have that zip file

[knotty moth]

or, for your current one, try downgrading gradio
env\python.exe -m pip install -U gradio==5.14.0

[knotty moth]

not in that way, my dud

[elder coral]

worked

it still doesn't work

that means i need a big pc

[knotty moth]

gradio is still fucked up?

[elder coral]

yes

[quiet axle]

anybody got a dual pc setup guide for w okada VC? the one on the rentry thread is pretty confusing

[pastel oak]

Dont think theres any other guide aside from the github, which is the same thing on rentry guide

Feel free to make a #1192011222023950368 and explain the steps youre stuck on

[pastel oak]

Which virtual cable are you using and is it a headset with a mic attached? And send a screenshot of wokada

[wraith light]

มะ

[low shard]

this server is english only

[somber canyon]

What Chunk setting and Extra should i use, or is recommended? i Got a 3060 with 12GB VRAM

[fiery juniper]

uh idk if this is the right channel to ask this, but is the W okada fork faster than the latest w okada?

[somber canyon]

what is fork?

[fiery juniper]

the uh

modified w okada thing

[somber canyon]

is there any vids about it? i never heard of it before

[pastel oak]

Yes

[fiery juniper]

o

[pastel oak]

https://rentry.co/ForkVoiceChangerGuide

[somber canyon]

is fork a local thingy?

[pastel oak]

Yes

Same as wokada just optimized

[somber canyon]

how do i know if i have it?

[pastel oak]

Runs on webui instead of an app to be less bloated

[pastel oak]

If you never seen the guide i linked you wont have it probably

[somber canyon]

i saw it before

[pastel oak]

Theres some small changes

[somber canyon]

i have the one where it opens a site in the browser

[pastel oak]

Does it say something with b2xxx numbers on the top right

[somber canyon]

yeah

[pastel oak]

Thats the fork then

[somber canyon]

oh cool

is there any recommended settings i could use

[pastel oak]

Whats ur gpu

[somber canyon]

.

3060 with 12GB VRAM

[pastel oak]

Start with chunk 192 and extra 2.7
F0 det rmvpe

Go to advanced settings and enable forcr fp32 mode on

Crossfade length:
0.05 faster but less quality
0.1 balanced
0.15 slower but better quality

---

For your audio inputs and outputs, select server mode, s.r. 48000 and use [windows wasapi] as a prefix on evdrything. This cuts down more delay than client mode but cant use the boxes like sup2

After that technically you can lower chunk as close as to the perf number as you can but never below it, but make sure it green. Thats like a performance indicator in how well your gpu is doing for the resources used (eg if u play a video game it can increase)

But 192 chunk is a safe value and gives more time for voice to sound clearer

[somber canyon]

cant i use client?

??

[knotty moth]

it has more audio delay but allows using Sup2

[somber canyon]

what is sup 2?

[low shard]

better noise suppression option

[simple ore]

try upgrading instead... 5.23.1

[simple ore]

see ^

or just download the compiled version and simply unzip

unless you desperately need something from the main branch and not 5.2.8-bugfix

[deft pewter]

is anyone having troubles with w-okada ?

it keeps on putting volume_in at 0 aft a while

and then i cant speak

unless i change the audio of the input device then switch back to my microphone i can speak but it just happens again

i can agree this didn't happe nbefore to me

[low shard]

share a screenshot of ur wokada

!give-media-perms 1h @deft pewter

I hope u aren't using those old ass youtube tuts for wokada

[deft pewter]

no?

i've been using it for more than 4 days

and it worked fine

just this recent issue

[low shard]

share a screen of ur entire wokada

[deft pewter]

i have to keep switching between this

[low shard]

yep ur using original wokada

original wokada doesn't have the perf value

also

[deft pewter]

im using the cuda 2.0.76?

[low shard]

vb audio cable also is reported to having random issues on wokada

[deft pewter]

it doesnt matter about vb audio cable

thats something else rn and its perfectly fine

[low shard]

it does, along with your wokada version

vb audio cable randomly stops working

[deft pewter]

how?

[low shard]

it's been reported by other users who said the same as you

it's a vb audio cable issues on windows

[deft pewter]

ok but im using my mon as my speakers rn

and its still happening

without even using vb cable

[low shard]

you should uninstall that wokada version along with vb audio cable

-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]

there's this new wokada deiteris fork which is an improved modified version

we suggest users to use that instead

[deft pewter]

oh

well here

basically if i have my mic open

and i switch between my mic

it keeps working fine untill i just disable the mic

this never happened yesterday or before lol

[low shard]

it's better you uninstall vb audio and that wokada version and try the deiteris fork guide

[deft pewter]

well

is there any alternatives to vb audio?

[low shard]

yes vac lite, which is explained in the deiteris fork guide

[deft pewter]

ah

[low shard]

did you firstly use a youtube tutorial to install the wokada you got?

[deft pewter]

wait is like

[low shard]

because all video tutorials are old

[deft pewter]

noo dude

i used wokada like wayy before

i just rejoined this community recently

and i haven't touched the settings a bit

[low shard]

yep, the wokada we used wayy before is pretty old

[deft pewter]

considering brandly reinstalled it

[low shard]

things change

[deft pewter]

well\

[low shard]

AI progresses ALOT overtime

[deft pewter]

is the deiteris fork

an entire

voice changer fork of wokada

or just a guide fork

[low shard]

it's a modified version of wokada, so a program too not just a guide

[deft pewter]

oooh

ily dude your my new fav jr admin on here fr

[low shard]

we always provide to the latest versions, keep in mind that the versions said online aren't as much up to date since the wokada and rvc trend died

[sly furnace]

how can you make your mic sound bad so people cant tell its a voice changer?

[deft pewter]

just focus on the pitch, volume

[sly furnace]

i did , its too clear people say its fake

[deft pewter]

does it like

cut off

[sly furnace]

some admin suggested me to make your mic bad

[deft pewter]

too fast?

[sly furnace]

no no everything is perfect buddy

thats the only problem

[deft pewter]

hm

[sly furnace]

it looks fake

[deft pewter]

what volume do u have it at?

[sly furnace]

400 💀

[deft pewter]

WHAT

[low shard]

-# I'm the only junior admin lmao

[sly furnace]

the 100 sounds too low to me

[deft pewter]

bro i usually set my mon or volume output to 0.8

[sly furnace]

or maybe i am deaf haha

[deft pewter]

or even 0.6 and it sounds perfect enough

[sly furnace]

0.6 wdym

wait

[low shard]

could u a share a screenshot of ur wokada

[sly furnace]

year sir one sec

[low shard]

!give-media-perms 1h @sly furnace

[deft pewter]

bro is using the ancient relics wokada 💔

[sly furnace]

maybe i am a bit confused

[sly furnace]

no no buddy i am using the fork

[deft pewter]

o

[sly furnace]

wait sir its loading

[deft pewter]

evelynn

[low shard]

alr

[deft pewter]

if ur settings are already good and stuff

its prob just the model

[deft pewter]

i usually use monitor instead of output

[sly furnace]

why so buddy ?

[low shard]

try lowering the extra to 2.0 and also to uncheck sup2 so it sounds more noisy

[deft pewter]

idk its basically the same but i feel there are quite some changes when i did it

[low shard]

also 400% is too loud

[deft pewter]

but in my opinion it feels good ngl

[sly furnace]

but it takes my keyboard and breathing sounds

it becomes too sensitive

[deft pewter]

legit true <<

[low shard]

output and monitor are 2 different things

monitor is for hearing your self

[sly furnace]

it sounds good to everyone , i also thought the same in starting

[low shard]

output is needed for where the converted voice goes

[sly furnace]

i lowered the extra to 2 and turned off sup 2

now what to do with my keyboard

and breathing noise

[low shard]

i feel like you tried to hear yourself without using monitor alone and basically had to put the volume out to 400 and the headphones made echo 😭

[deft pewter]

well there isn't really anything different with it right? u can use them both either way as ur output of audio virtual cable

[low shard]

what if you try to put the microphone a bit far ? if it doesn't solve anything then nvm leave it on

[sly furnace]

yes i did it with hear myself on discord , and also about the other thing , when i use my mic and output of my headphone , the mic captures desktop audio as well so it converts the sounds of the people talking to me too , so i fixed this by using my headphone mic and using IEMs as output

[sly furnace]

i use headphone mic its stuck to my face but it still captures my clicky keyboard sounds idk why

[low shard]

you can lower the volume of ur headphones to fix the issue of it capturing the desktop and other people talking to you

[deft pewter]

yeaa

[sly furnace]

yes i did it works but i have to lower it too much so i used other output source instead

[deft pewter]

i dont think ppl would wanna hear their own voices in a ai voice 💔 i usually just use a microphone and speakers instead of headphones

i also use those microphones that are configurable too

[sly furnace]

now when i am running it , my dad is in another room talking on phone to someone its also capturing that audio 😭

thats why i use noise supression on

what should i do sir ?

[low shard]

it's better to use output just fo re-routing the audio and monitor to hear urself

[low shard]

check sup2, also lower the audio of ur headphones

the issue is that the model still sounds too good?

[knotty moth]

monitor is to listen the output voice, the output one as virtual cable will forward to another application using it

[sly furnace]

can i show u in vc or something so u can tell better?

obv if you are not busy

[low shard]

I can't really VC

[deft pewter]

a question rq

how do i use

the normal app version

[sly furnace]

one sec

[deft pewter]

i dont wanna use the website version of this fork

💔

[sly furnace]

ill record and send

[sly furnace]

why so

[low shard]

there isn't

both programs always used a web user interface

the only difference is that the original made it's own "browser" to open the web user interface

[deft pewter]

how does this dude

have the blue bars

[sly furnace]

i am running the fork

[deft pewter]

oh

having the blue bars must be a flex 💔

[sly furnace]

why so

😭

[deft pewter]

wanna see why

lemme get ss rq

[low shard]

the great majority of AI runs on a web user interface

[sly furnace]

i cant send voice msg here

[low shard]

like 90% of all ai, because it's way easier to edit which is needed for programs that always change and progress

[sly furnace]

give me perm

[simple ore]

HiDream

[deft pewter]

💔

in despair rn

[sly furnace]

😭

[low shard]

@deft pewter btw the program still runs on your own hardware, it's just the interface running on local web

[low shard]

what browser are u using btw

u need to set the gpu to ur gtx

f0 to rmvpe without onnx

[deft pewter]

wait

bro

i set it to my gpu

and also the issue is happening again bruh

[sly furnace]

help me

first

[deft pewter]

the mic issue is why this is happening

[sly furnace]

why it sounds fake

😭

[low shard]

you didn't even set the audio settings nor are using the microphone rn

[pastel oak]

probably bad model

[low shard]

have you tried with other models?

[sly furnace]

give me good model than ?

which is the best one

[pastel oak]

theres models recommended on the guide

[sly furnace]

i tried them

[low shard]

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

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

[sly furnace]

i dont want to disturb u guys with my stupidity so after trying everything

i came to you

i know your time is valuable

[low shard]

it's better you test other models too

[sly furnace]

i tried them sir

[low shard]

^

i gave you also other ways to find models

does it sound too low quality or too good quality?

[sly furnace]

alright sir give me some time ill try and let you know the results

thank you for the info

[low shard]

gpu: ur gtx
f0: rmvpe without onnx
input: microphone
output: line 1
monitor: optional to hear uself, set to headphones

[deft pewter]

that photo was just first preview

[low shard]

if u still have an issue, please elaborate

[deft pewter]

i already did it all

just that the microphone issue is following me everywhere rn fr

just on wokada

i think

[low shard]

could you explain exactly what you mean by that?

[deft pewter]

oh my

where the hell is my

input device in sounds setting??

its gone 💔

[low shard]

are you doing the device settings fix in the 3rd step of the guide ?

[deft pewter]

didnt see that mb

[low shard]

u need to else it will fuck things up

also uninstall vb cable

[deft pewter]

this right

done

[low shard]

yes

[deft pewter]

also

where is the 3rd step

i cant find it

[low shard]

https://rentry.co/forkvoicechangerguide#virtual-audio-cable

[deft pewter]

what do i do bruh

the shemboy mic

is legit gone 💔

even tho i have it plugged rn

[low shard]

please follow that 3rd step and then reboot your windows

you need to also follow this, it's important in the 3rd step

[deft pewter]

thats what im doing bro

just the microphone im trying to set it to

is gone

from the list

[low shard]

it isn't in mmsys.cpl either?

[deft pewter]

yea

[low shard]

if not, restart your pc and check if it's there

[deft pewter]

it aint in it

i restarted my pc already

[low shard]

are you sure it isn't that Microphone USB Audio Device?

have you tried it?

[deft pewter]

no i remember it legit had a name

in mmsys.cpl

[low shard]

could you try using that and check if it works?

[deft pewter]

alr?

nope not it

thats just my speaker

[low shard]

weird as it's called microphone
could you please check if there's any windows update

and if not
search on windows device manager and check the input/output devices

[deft pewter]

no updates

huh

its also gone from there

it was working perfectly fine this morning??

[low shard]

did it disappear only after installing vac lite?

[deft pewter]

pretty sure yea

mhm

it was there in old wokada when u told me to install vac lite the entire device just disappeared

[low shard]

could you try uninstalling it, check if the microphone is back and also try reinstalling it

[deft pewter]

ok

wait

how do i uninstall it

found it

ok now its not even back anymore

i think vac lite casually

just deleted the entire whole device

ima restart

uh

now its back

it only came back aft i deleted vac lite ima reinstall and see now

ima test now

uh now shem boy mic is also gone again

@low shard pretty sure

that vac lite is just deleting my mic when i have it installed

yea

ima just another virtual cable

[low shard]

this is pretty weird as I never seen it happen before, @pastel oak have you ever encountered this issue?

[deft pewter]

uh

@low shard

turns out the microphone cable is actually getting old

so now even if its connected its connected just that not fully

to a point on where if i move the cable my self the pc might just recognize it and if i stabilize it

it stays

bruh i didnt know my stupid microphone cable got old

[low shard]

welp things get old

[trim sparrow]

anyone know how to fix this “ Traceback (most recent call last): File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\infer-web.py", line 292, in vc_single audio = load_audio(input_audio_path0, 16000, DoFormant, Quefrency, Timbre) File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\my_utils.py", line 103, in load_audio raise RuntimeError(f"Failed to load audio: {e}") RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail) Traceback (most recent call last): File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\routes.py", line 437, in run_predict output = await app.get_blocks().process_api( File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1349, in process_api data = self.postprocess_data(fn_index, result["prediction"], state) File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1283, in postprocess_data prediction_value = block.postprocess(prediction_value) File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 2586, in postprocess file_path = self.audio_to_temp_file( File "C:\Users\Joshu\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 360, in audio_to_temp_file temp_dir = Path(dir) / self.hash_bytes(data.tobytes()) AttributeError: 'NoneType' object has no attribute 'tobytes'"

[quasi condor]

https://cdn.discordapp.com/attachments/1352588755214794764/1358966424596644022/image0.gif

[trim sparrow]

thanks for the help🙏

[crude flame]

Light host my beloved

[sly furnace]

what does that mean sir?

[golden walrus]

guys, can i ask how flatline look like ?

[sly furnace]

voicemeeter?

[crude flame]

A vst host

Let’s you apply vsts to your mic

[sly furnace]

will that make my voice sound real ?

https://www.youtube.com/watch?v=AyzQZScP2Sw

is this the exact thing sir?

do u have better tutorial ? (sry to disturb u )

[crude flame]

Depends on what you mean by “real”

[sly furnace]

the current voice model doesnt sounds real

i want it to sound real

[knotty moth]

define your "real"

[sly furnace]

real means sound from a real person , like people detect its ai

i am sry if i am sounding stupid

[knotty moth]

"real is real" is not a correct definition

[sly furnace]

when i use the voice changer , people can tell easily it sounds like ai

i want to fix that

[crude flame]

Does the model sound to artificial

[sly furnace]

i am currently using w okada fork

[sly furnace]

yes we can say that

it sounds like a girl but has some issues

i tried 2-3 models

[crude flame]

If it sounds like a robot then it’s a model problem

[sly furnace]

i know i might be doing something , i just want to know what is it

[sly furnace]

it doesnt sound like robot sir

[knotty moth]

just say something specific like "robotic" sibilants or some distortions

[crude flame]

Have you tried acting? Like try to also talk like the voice so it sounds more real

[brittle wing]

I wanna start training my model next week, how do I resume on Applio no UI?

[crude flame]

So it sounds to clean? Has that rvc tone to it?

[sly furnace]

uh its hard to explain 😭

[sly furnace]

yesssss

its cleannn

i want it sound like real person , like real persons have immperfections

little iff buts

[knotty moth]

just play some background noise/ambience also lighthost with some vst plugins

[sly furnace]

can u guide me on that ? if u can provide me any tutorial ? i cant find a good one on youtube

[crude flame]

The you can install light host and then apply a filter and bitcrush and maybe apply some bg noise like a fan

[sly furnace]

i can try that

can u send me a guide if u have?

[crude flame]

It’s an old guide I made but it still works (I think)

https://rentry.co/W-okadaVoicemeeter

[sly furnace]

alright sir i try it out ? do you mind if i ping you if i have questions ?

[crude flame]

You can ping me if you have any questions

[sly furnace]

Thanks a lot sir for helping me out ! i really appreciate that

[low shard]

Mangio RVC is extremely outdated

Never use that

[trim sparrow]

what should i use

i just found it on youtube

[low shard]

If u followed video tutorials, they are all old, only written guides are updated

[low shard]

Yup all YouTube tuts are all old for rvc

[low shard]

What's your PC GPU and what do u want to do

Mangio RVC is abandoned since 2023 for example

[trim sparrow]

1650 super and i have a friend who wants to sing and i produce and engineer for him but he sounds like shit so i wanted to see if that would fix his vocals and just to mess around with the vocal models

[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.gg & rvc-ai-cover-maker-ui
easiest cloud: Ilaria rvc zero
easiest local: Applio

[trim sparrow]

do you think weights.com or applio would be good for what im trying to do?

[low shard]

Yea, the quality depends mostly on the model btw

[trim sparrow]

are they free?

[low shard]

Yes, Applio is open source and has no limits unless you use cloud (because that depends on the Cloud service provider), weights is much easier to use but it has free limits (since it's cloud, and they let you use their own GPUs), however the limits are pretty good

[trim sparrow]

sounds good thank you so much ima retry mangio js rq because since i downloaded it already but if it doesnt work ill look at applio thank you🙏

[low shard]

I wouldn't suggest mangio at all

Full of bugs and outdated asf

It's maintained on hopes and dreams

[trim sparrow]

yeah ive been trying for a min n its just being stupid ima look at applio (downloading it rn)

[gray cobalt]

How do I delete models, just deleting their files is enough?

[low shard]

Which program are you talking about?

[gray cobalt]

Deiteris w-okada fork or something

[urban fractal]

Hey guys, just joined, awesome group. Quick questions, I'm using python with tts_rvc like the following but I can't seem to get her voice to work, he sounds russian. I'm not sure what to set the voice to when I use the model path.
Any suggestions?

tts = TTS_RVC(
model_path="ARIANAGRANDE_ES_BY_SZAJEAN.pth",
index_path="added_ARIANAGRANDE_ES_BY_SZAJEAN_v2.index",
f0_method="rmvpe"
)

[low shard]

Oh Wokada deiteris fork, you can just overwrite the slot

[gray cobalt]

Ah

[low shard]

Or deleting the model files should be enough

[gray cobalt]

Thanks

[low shard]

Yw

[urban fractal]

FYI: Im using this github: Maybe this is the issue?
https://github.com/Atm4x/tts-with-rvc