✨│ai-help

[low shard]

can you elaborate what i asked and what u want to do?

[latent cypress]

i finally got my cmd to stop running as admin,
and now i'm getting this problem when running run-applio.bat

[low shard]

Applio-main? Did you get the precompiled 3.2.9?

which gpu do you have btw?

[latent cypress]

i downloaded it from here

[latent cypress]

rtx 4060

[low shard]

is there a specific reason why you're trying to get the main branch? it might have changes that aren't stable yet, it's suggested to follow https://docs.aihub.gg/rvc/local/applio/#nvidia-on-windows-precompiled

[latent cypress]

i was looking for the normalization option in the dataset section,
lyrey said "use the main branch, not the compiled version"

[low shard]

Ah, I see
Did you place that folder in a C drive, without priviligized access, and without special characters?

[latent cypress]

there's always some issue with my dataset, and my grads norm is always above 500.
some people even said my grads norm is way too high at the start

[latent cypress]

yes

C:\Applio-main (my applio path)

[low shard]

great, you an the bats without running them as admin, right?

[latent cypress]

yep

[low shard]

could you please reinstall again? i just tried again and should work all fine

[latent cypress]

okay

[golden walrus]

guys, anyone had run into UVR UI couldn't download, no permission ?

i have no idea why this is the problem now

[quasi condor]

uvr5 is broken

[golden walrus]

uvr UI

not uvr5 lah

[quasi condor]

same thing

lah

[golden walrus]

i thought local uvr won't be affect

[quasi condor]

well u can use uvr5 :p

[golden walrus]

but i don't know how to install model

[quasi condor]

or uvr5 ui colab if u have a potato pc like me

[latent cypress]

it worked, now it's downloading the files

[low shard]

the project by @viscid moss is called uvr5 ui , even tho i understand that can create confusion

[low shard]

local uvr isn't affected

[low shard]

what's your pc gpu? could you explain what you did and a screenshot?

[low shard]

great then!

[golden walrus]

i mean, i can't download the seperated audio

[latent cypress]

when i tried deleting applio earlier, there was one file that needed admin rights to remove. not sure what the full name was. but thanks, it works fine now

[quasi condor]

i didn't know that exists

also WHY IS IT ON LIGHT MODE

[quasi condor]

https://tenor.com/view/ash-baby-gif-7280845142657635823

[low shard]

ohh you're using eddy uvr5 ui locally, this seems an issue in microsoft edge, please check if any anti virus are giving issue, try checking microsoft edge permissions, and also try other browsers

could also check https://www.youtube.com/watch?v=KmOpgvyRZgQ, but it seems a browser issue rather than program

[viscid moss]

Wtf

@golden walrus results are into /outputs folder

No need to re-download

[golden walrus]

oh god

oh godddddddddddd

until this day

thank youuuuuuuuuu

[viscid moss]

ur welcome

[hallow thistle]

This is the whole new W-Okada I extracted and installed from start because I deleted a partition of this hard drive by accident. Bonus: Windows Terminal from Windows 11.

[viral mason]

It's so bright agghg

💔

[hallow thistle]

That's an issue with your eyes.

[low shard]

everyone prefers dark mode tbh

[hallow thistle]

Because some bad people forced me to use dark mode, here, I have already set Discord to dark mode just so to satisfy you all.

[viral mason]

I feel bad for light mode users, on anything

[hallow thistle]

Seriously, don't feel worry about me. I'm in bright place. You are whatever you are.

[viral mason]

What

[hallow thistle]

Automatic1111 crashing because ran out of memory, until I realized it attempted to use a stock SD 1.5 checkpoint model, which is over 3 GB, my laptop RAM won't gonna handle that.

[chrome wave]

Hi, may anyone please help me with w-okada voice changer. I've used it only once before on my friend's PC, but I'm an AMD user and I cannot seem to find the download for AMD

What

[low shard]

Hello, could you please elaborate?:

• your pc gpu
• operating system
• what you want to do:
  • ai covers
  • tts
  • e girl trolling
  • roleplay in vc
  • roleplay in games
• the tutorial link or screenshot of the entire program

[small garden]

Hello

[chrome wave]

Well my friend uses it for art but she couldn't provide me with the file because she's nvidia user. I'd just need it to configure my voice, possibly some pitch etc... depends

she said w-okada is best for that

[small garden]

One question, how did Google clone the voice of Google Translate when there was no good TTS at the time like Tacotron 2?

[low shard]

this program isn't related to art at all, are you looking to do art? or ai covers? or e girl trolling?
There are different programs for different things, I need to know what thing you want to do to give you the best program

[small garden]

questions that have no answers

[chrome wave]

oh no we are actually doing ART, we are looking to use it for AI covers (especially youtube video intros)

[analog obsidian]

if your dataset is clean and nothing is wrong with it, then probably you're getting high grads because the dataset is very different from that of the pretrain

[small garden]

I know the name of the voice conversion programs

can I talk the name

[chrome wave]

@low shard is there an download link, I cannot seem to find it?

[low shard]

then it's not the right program

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. There also updated forks with extra features like Applio.

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)

which one do you actually want?

[small garden]

You know?

Hello ffdf

[chrome wave]

Wokada, the one you said that uses RVC (at least that's what my friend told me to get)

The deiteris fork

[latent cypress]

what do you mean by different? my dataset’s all singing audio, btw

[low shard]

wokada and rvc are 2 different things, wokada deiteris fork doesn't do ai covers

[small garden]

Hello

[analog obsidian]

well the og pretrain dataset is all monotone speech

[chrome wave]

oh she said deiteris fork, sorry im not really familiar with this.

[small garden]

Do you know what the first voice conversion in the history of the internet was

[low shard]

soo, please elaborate again what you want to do, you seem to be confusing the programs, please tell me exactly what you want to do so i can suggest the best program

and also tell your pc gpu and operating system

[latent cypress]

https://cdn.discordapp.com/attachments/1160667557389402193/1331871154822053901/br.gif

[analog obsidian]

could be also the dataset quality might be too low many separation models used maybe?

[small garden]

it was talknet

The first voice conversion

[chrome wave]

Well she originally does youtube intros with her own voice, but she's using custom pitch settings etc... to make it better
She wants me to help her, we want to do it together

my GPU is amd radeon 6700xt, windows 11

[small garden]

but he died in 2023, F for Talknet

[low shard]

so, use models on pre-recorded audios right?

[chrome wave]

Yes, she's doing youtube intros for youtubers

[small garden]

a question which is better SVC, RVC OR GPT-SO-VITS?

for me to use

[low shard]

You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides):
  • Applio (AMD Windows) : A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
  • Mainline (AMD Linux/Windows) : 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 Colab: max 4 hours daily, not granted, of GPU
  • RVC-AI-Cover-Maker-U Colab & Kaggle: 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 colab/kaggle
easiest cloud: Ilaria rvc zero
easiest local: Applio

[low shard]

those programs aren't even related
So-VITS-SVC and RVC are Speech To Speech (STS), you should use rvc
GPT-So-VITS is TTS

[chrome wave]

Just to make sure, which one is the "deiteris fork"

[low shard]

none, wokada deiteris fork is the wrong program, wokada deiteris fork isn't meant for pre-recorded audios it's only realtime for roleplay

[small garden]

and talknet that you need to write the lyrics of the song to make the voice sing,

[low shard]

you and your friend should use RVC instead

[chrome wave]

but she's asking me to get the deiteris fork, is there no way to get it or?

[small garden]

as if it were a Vocaloid

[low shard]

i'm not sure about talknet

[low shard]

it's the wrong program, yall are confusing programs, RVC is for ai covers and pre-recorded audios

wokada deiteris fork is only for roleplay

[small garden]

I've already used it, it was on Uberduck

but the owner of uberduck was sued, and had to remove the voices

[chrome wave]

No I do not im serious but very confused about this, she's saying the deiteris fork is the right one and you're not sending me the link to it

[low shard]

deiteris fork isn't for ai covers 😭

simply, she confused the programs

it would be better you tell her to join here too

[latent cypress]

3 (voc fv4, karaoke gabox, and dereverb mono)

[small garden]

owners of the TTS sites, FakeYou:echelon, Uberduck:ZWF, Falatron:Cris140, 15 ai: 15

I know about TTS sites

[errant basin]

What is the best configuration for a 3060? Female voice? Without the voice taking too long

[low shard]

What is the best configuration for a 3060?
you prob came off youtube, they are all outdated, delete original wokada folder zip and uninstall vb audio cable from windows app settings
Female voice
Are you looking to do e girl trolling?
The new programs depends on your use case, there isn't a program that does everything, so it depends if you want to do like e girl trolling or tts or ai covers

[errant basin]

I just downloaded the new Wokada to use in Red Dead roleplay, but the voice has a very long delay

[low shard]

Please Elaborate:

• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • E Girl Trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[analog obsidian]

[latent cypress]

btw some of the audio clips in my dataset are over 3 seconds, is that gonna be a problem when applio processes it?

[analog obsidian]

nope

applio/rvc adds a small overlap at the start of each slice to preserve context

[latent cypress]

ooo okay okay

[latent kettle]

@simple ore how to train model on rental GPU.

[brittle wing]

@simple ore yo vast.ai support replied

they said this

Hi thanks for your interest in using vast.

To train an RVC (Retrieval-Based Voice Conversion) model on a rented GPU, here’s what you’ll need to do:

Choose the Right Template:
We recommend using a Linux-based template with CUDA support. You can use the Ubuntu 22.04 image (ubuntu:22.04) or a pre-configured AI/ML template like pytorch/pytorch or nvidia/cuda:12.6-devel.

SSH into Your Instance:

Can follow this here: https://docs.vast.ai/instances/sshscp

Install Required Tools:
You can install Python packages directly using pip. For example:
sudo apt update && sudo apt install -y git python3 python3-pip
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126

Clone the RVC Repository:
git clone https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI.git
cd Retrieval-based-Voice-Conversion-WebUI

Install Dependencies:
Run the required setup:
pip install -r requirements.txt

Upload Your Training Data:
You can use scp or rsync to upload your dataset to the instance
Start Training:
Follow the RVC guide to configure the model and start training. Most UIs allow you to point to your dataset folder.

If you're using a template with a web UI (like Jupyter or a custom Flask app), you may also upload files directly via the browser interface, but SCP is more reliable for large datasets. Also if you are just getting started on vast check this out: https://docs.vast.ai/quickstart

[simple ore]

yeah, pretty much that, get ubunty instance with nvidia support, install python 3.11, clone applio repo, install requirements

you then move the prepared files into logs

can run the training script right from the command line

[brittle wing]

how can i do all that 😭

like i know how to install python and the requirements and stuff but idk what ubunty instance is

[simple ore]

I've never used their service, so I dont know the exact steps, so as they explained, you need to pick a preconfigured AI/ML template with CUDA

then you ssh into the instance, install python if it is not installed/wrong version

then you clone applio

then you run pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126

then you run pip3 install -r requirements.txt

use WinSCP to copy files over

[latent kettle]

Do you know any other alternative which you have tried

[simple ore]

I used a plain linux install

which is not all that different from that

[brittle wing]

bet thank you

[undone orbit]

Evening chaps, is v5.6 - UVR GUI the best vocal remover I can get at the moment ? I want to run it locally. Want to use to extract vocals from rap songs.

[simple ore]

it is quite outdated

Eddy's versuion has new models and stuff

https://github.com/Eddycrack864/UVR5-UI/releases/tag/v1.8.4

download precompiled version

[brittle wing]

@simple ore these are the main ones that i can rent bro do u know which one i need to use to train on the applio thing?

cause the guy said to get a cuda but you mentioned pytorch as well

[simple ore]

nvidia cuda

I think there's little difference in what is installed by default

[brittle wing]

aight

[shrewd plover]

i have rtx 5080 win11 @low shard

[low shard]

Please Elaborate:

• what you want to do?:
  • AI Covers
  • TTS
  • E Girl Trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program (if any)

[viral mason]

this shit broken

https://huggingface.co/spaces/qtzmusic/UVR5_UI
it doesn't even work when I insert audio into it and just spits out an error saying I exceeded the limit even tho I haven't

[low shard]

hf uvr is being half broken, just use an alternative

@viscid moss

[viral mason]

I'll switch to the collab till this gets fixed👍

[sour sun]

Hey again, may get back into this soon

Quick question about dataset variety

I know that it's recommended to remove repeated words and sounds from a dataset. My question is: if a word is repeated in a different manner, is that counted as variety or should it be pruned?

Ex: character says "apple" three times, all in neutral tones, but slightly different intonation and pitch. Should I prune until one is left?

[low shard]

till this gets fixed👍
it's not something they can fix, it's an unknown rare zerogpu bug, all they can do is contact hf and hope it works

[viral mason]

well at least there are other options than just huggingface for now

thx for at least looking into it

[terse scaffold]

Hey is there a Google collab or something like that so I can make models without having to download anything?

[low shard]

there is, but it shouldn't be your first option, it has limited time, you should first check if your pc gpu is good enough to do it locally

[terse scaffold]

it's not
that's why I'm asking

[low shard]

Train (make) RVC Models on cloud:

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

• Google Colabs (max 4 hours of daily T4 16gb gpu not granted for free, not much hours for training, but easy to use, there's a paid tier):
  • Applio (UI)
  • Mainline (U, END OF LIFE UPDATE)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus, either T4x2 16gb each or P100 16gb, only free):
  • Applio (UI)
  • Mainline (UI, END OF LIFE UPDATE)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly, Free Studios run 24/7 but require restart every 4 hours. There's a paid tier):
  • 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

[terse scaffold]

Aighy thx

[viral mason]

I recommend Kaggle

applio

it's pretty good

[timid dune]

@low shard

how do i get ping down again

nvm

[terse scaffold]

It's not working

I followed the guide and everything

[low shard]

aòsp id suggest applio kaggle for trainign for more free gpu time

[terse scaffold]

Aighy

I'll try it

[terse scaffold]

It doesn't explain how I'm supposed to login in File Url

[low shard]

wdym? can you share a screenshot of the issue?

[terse scaffold]

this thing

I used the kaggle one

applio

[low shard]

can you try putting "applio" as username (leaving password blank) then login?

[terse scaffold]

will try

[terse scaffold]

Not working

says error 500

[low shard]

what if you refresh, and login without adding any username nor password?

not sure if it got changed recently, will test rq

[terse scaffold]

tried
same thing

man
what happened to the old RVC disconnect Google Colab thing
It was so easy to use
Everything in the same link
didn't have to go right and left just to make a silly model 💀🙏

[low shard]

starting the notebook rn and checking myself soon

[low shard]

the old rvc disconnected colab will never comeback, i mean it had issues with 2023 outdated code, and the creator doesn't use rvc in months if not a year now

do you want the easiest possible option ever to train RVC models in cloud? That's https://weights.com/models

[terse scaffold]

sad

[low shard]

it's all automatic, if you really don't want to mess with RVC yourself, since weights.com is 1 click

[low shard]

it's actually a good thing, i used that too back in the days, but the code would be severly outdated now, it used a mangio rvc fork abandoned since 2023 lol, messy with worse performance

[terse scaffold]

I guess

[fleet cedar]

is there anyway to use line 1 as a AI TTS for vc?

?

[low shard]

@simple ore there seems to be an issue with applio kaggle
setting filebrowser user applio with admin perms, but without password, breaks it up, but you can just replace the blank password in the code with "applio123456" and the users won't even get the login screen, this is because the no auth thing is related to the user with the 1st id
Not sure if this is a bug related to a newer version of filebrowser not accepting blank passwords

edit: i made a PR https://github.com/IAHispano/Applio/pull/1100

@terse scaffold just change the blank next to applio, to "applio123456"

[low shard]

https://docs.aihub.gg/tts/realtime-tts/

[terse scaffold]

should I rerun it afterwards?

[low shard]

it would be better you factory reset everything, change that line and run everything again, should be all fine

seems to be related to a bug with a newer version of filebrowser

you can also use weights.com if you rather a 1 click easy option btw

[terse scaffold]

Aighy
Will let you know if it works this time

[terse scaffold]

ye ye
still using it in the meantime

[terse scaffold]

okay it works now

doesn't ask for login

thanks

[viscid moss]

@viral mason problem is request time on that one is "600" that's too much. Ilaria's one is just 60. Maybe try this one (there's 300):
https://huggingface.co/spaces/Floofmusic/STATION

[fluid epoch]

im getting this help : C:\Users\avyxl\Downloads\vc\MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a\MMVCServerSIO>MMVCServerSIO.exe -p 18888 --https false --content_vec_500 pretrain/checkpoint_best_legacy_500.pt --content_vec_500_onnx pretrain/content_vec_500.onnx --content_vec_500_onnx_on true --hubert_base pretrain/hubert_base.pt --hubert_base_jp pretrain/rinna_hubert_base_jp.pt --hubert_soft pretrain/hubert/hubert-soft-0d54a1f4.pt --nsf_hifigan pretrain/nsf_hifigan/model --crepe_onnx_full pretrain/crepe_onnx_full.onnx --crepe_onnx_tiny pretrain/crepe_onnx_tiny.onnx --rmvpe pretrain/rmvpe.pt --model_dir model_dir --samples samples.json

[viscid moss]

I'll try a fix for Ilaria's one that someone recommend me

[fluid epoch]

hey can u help

[fleet cedar]

do i need to pay if i use local like subscription stuff from free to plus or smthing

[low shard]

you're welcome

[low shard]

subscription for what? which program are you talking about?

[fleet cedar]

fish speech

[low shard]

replied in https://discord.com/channels/1159260121998827560/1400624193481216060

[low shard]

if you use local, it's all free and open source

if you use the site, then there's a paid tier, your choice

[fleet cedar]

OMG teach me now

PLEASE

how do i install it bruh

like on local

[low shard]

what's your pc gpu?

[fleet cedar]

5070

idk if thats bad for TTS

😭

[low shard]

should be fine, there isn't a step by step guide on our docs and i haven't tried it locally, but you can try checking their github http://github.com/fishaudio/fish-speech

[fleet cedar]

ts is confsuing

[low shard]

yeah, AI is intensive and complex

try reading their Documents

i see their documentation is mostly related to linux, maybe try checking https://github.com/AnyaCoder/fish-speech-gui or asking @simple ore since i'm guessing you're on windows

[simple ore]

you need to install it

[low shard]

nope

might work ONLY for HF pro (paid), or not at all, not sure

[viscid moss]

Yeah, Pro users are fine but free user aren't

In that case there's no working HF space for UVR 5 UI

[low shard]

neither for ilaria rvc, still broken, some works, some not

[terse scaffold]

wait
so now hoe do I copy path from File browser to Applio
cuz it doesn't chose the option when I right click on the zip file

[viscid moss]

Yeah, too much time requested

[low shard]

just use cloud uvr or local 🙏

[viscid moss]

Or UVR5 UI locally

[low shard]

or... become ai and separate them urself

[viral mason]

local is a no for me I've explained why enough

[viscid moss]

XD

[viscid moss]

🫠

[viral mason]

guess Im stuck with the less organized collab version

[low shard]

sorry, could you elaborate the issue? what option?

[viscid moss]

Hmm Colab isn't hard to use

Just 2 steps

Install and run

[low shard]

yeah and might be for a while, if not forever

[viral mason]

it's not that it's complicated I just dislike the list since it's not organized very well like it is in the huggingface space

[terse scaffold]

to copy path
like how am I supposed to get the voice samples from File browser to Applio

[sour sun]

You can also select the item and click

[terse scaffold]

why don't I have that 💀🙏

[viscid moss]

but how, the inferface is like the same 😭

Just more features that i can't put on HF cuz ppl will broke the app

[viral mason]

it looks nothing like the hf space

this is not that

[viscid moss]

that's not my UVR5 UI Colab

lol

-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

[terse scaffold]

I even pressed the share button to see if it would work

it didn't

[viral mason]

my friend sent me that one, kinda thought it was yours lol

[viscid moss]

nah

[terse scaffold]

How do you use kaggle bro 🙏🙏🙏

[viral mason]

are you trying to train a model?

[terse scaffold]

ye

[viral mason]

on applio

[terse scaffold]

ye

like I'm at the file browser part

[viral mason]

ok so I have no idea what this is but watch this

to get to that training tab I was on in the video wait at the third cell until it pops up the first link then also click the second which is tensorboard

[low shard]

don't remove the branch part

it will use the main branch, which is experimental

not intended unless you're an advanced user

@terse scaffold you can just use applio's dataset maker

[viral mason]

I thought that was bad, if I keep it what would it change about the model I train?

[viral mason]

also this, is a guid for tensorboard

[low shard]

it is not "bad", it chooses which type of source code branch you will use

removing the 3.2.9, means you will use the main branch, which has more advanced and not confirmed stable yet changes
So yes you can use it, it could change the way to train (like add features or remove them or bugs), but it's not suggested to remove the stable branch unless you know what you're doing and like experimenting

[viral mason]

👍

I know what I'm doing with training so I'm going to continue doing that

[low shard]

alright, just don't suggest it for new users, might be confusing for them

but still thanks for trying to help users!

[viral mason]

np!

[analog obsidian]

the main branch usually have experimental shit that break stuff

just use 3.2.9

[simple ore]

current main is stable

if you want single mel default

[low shard]

oh lmao

[low shard]

btw if you could please that PR I sent you before

[terse scaffold]

mine said no dashboards are active for the current data set

[simple ore]

blaize did

[low shard]

Oh great!

[whole pecan]

how do i reduce the delay on vonovox fork?

[wraith nebula]

is there a tutorial on exporting a weights model to something like a voice changer? it doesnt seem to work, i have the .pth and .index file, but it doesnt work

[hallow thistle]

Voice models from Weights are RVC. They are basically the same as any other RVC voice models. But here's how you download a voice model from Weights.

[sudden sundial]

When Applio says 40000 Hz is it actually expecting 40000 or 44100?

I've only ever seen 32k, 44.1k, and 48k... 40k seems like an odd number

[hallow thistle]

More like 40000 Hz.

[sudden sundial]

Gotcha, thank you!

[hallow thistle]

From what I've seen, almost every RVC fork, even Applio, never has a 44100 Hz option for training. So, there are 32000, 40000 and 48000 instead.

[karmic steppe]

hey what should do for the girl voice changer. My voice pretty deep and when I try to up the tone it cracks alot

[naive pond]

Hey guys, it's been like a year since I've trained my own voice model, and I'm tryna get back into it. I'm following the Applio Collab guide, but once I get up to inference (after downloading the model link and refreshing under the Inference tab), there are no options for Voice Model or Index File in the drop downs. My link is a public google drive link of the training data as a wav file. Am I missing something?

[simple ore]

to train a model - upload the wav files using dataset creator on training tab, preprocess, extract features, train

after that refresh inference screen and the trained model will be available in the drop-downs

to download the model to your local pc, use the download option at the bottom of the training screen

[naive pond]

Thanks! Giving that a go now ❤️

[sudden sundial]

What do I set the GPU Number to in Applio?

[hallow thistle]

On Applio, in Training section, if GPU Number corresponds to your GPU name in GPU Information section; for example, if GPU Number says "0" and the GPU Information says "NVIDIA GeForce RTX 4090", it's the GPU number you're looking for. https://cdn.discordapp.com/attachments/1159290139609137264/1397602063067775058/image.png

Supposedly looks like this. https://cdn.discordapp.com/attachments/1159290139609137264/1397595090981224558/image.png?ex=688d801b&is=688c2e9b&hm=0c16acace334d3e626df7fb552967fa79e0e81c198f0ab59fac1800b33f5a330&

[sudden sundial]

Ohhh, thank you very much. I got confused for some reason that it might be like GPU "cores", but I couldn't (obviously) find an analog for cores in GPUs, so I didn't realize that it was just the number(s) of the GPU(s) you want to use

[hallow thistle]

Unlike CPU, no program would count how many core and thread of a GPU, so the number there would correspond on how many physical GPU(s) you have.

[sudden sundial]

Thank you

[blazing panther]

there a way to install nvidia w-okada on arch linux?

[hallow thistle]

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#download-on-nvidia-on-linux

[blazing panther]

The doc doesnt explain this version far enough, what should I do after downloading the files?

[hallow thistle]

These tar.gz files are splitted into two parts. If you know how to use terminal, use cat command in Linux to join them into one and extract them with tar command. Additionally, you can use 7-Zip Linux version or other known GUI ZIP/archive programs to extract them in single task.

[blazing panther]

I was already off setting up anaconda for this but if its that easy that works as well lmao.

[hallow thistle]

While Conda is not needed for W-Okada, it's definitely needed when you work around with files while you wanna use related tools from other directories within the same terminal environment.

Also, there's this one in the guide. https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#opening-on-linux

[terse scaffold]

I keep getting this error when I'm trying to train my model

"Not enough data present in the training set. Perhaps you forgot to slice the audio files in preprocess?"

Idk what's wrong I followed very instruction Someone help please. This is driving me crazy

Using kaggle Applio

[low shard]

are you trying to do e girl trolling or are you trans?

[low shard]

Perhaps you forgot to slice the audio files in preprocess?

[terse scaffold]

I did tho

[low shard]

you really sure you left this to automatic?

[terse scaffold]

yes

I didn't even touch anything from this option it was automatic and I kept it like that

[worldly terrace]

hi nick dont know if u remember me but iw as the guy with the weird problem

im using macbook now hope it works normally now

[fiery raptor]

Why do I keep getting this error? Its alr downloaded 💔
(this is the windows local vers of uvr5 btw)

[latent cypress]

applio sets fp32 as default right?

idk if i should care about grad_avg_50/norm_g, since it starts at 694, but grad/norm_g starts from 398

[pastel orbit]

i need help

with some settings

[flint vapor]

Hi, what happened? Elaborate your problem

With which software?

What?

[north barn]

in the rvc docs guide, what does the "perf number" mean?

[low shard]

it's the performance value at top left

be sure you get it from the ai hub docs guide, not youtube tuts btw

[north barn]

yep im looking at the docs guide now

thanks btw

[low shard]

great, want me to check your settings while we are at it?

[north barn]

no its fine i have it all setup properly, im just wondering what "perf" means

[low shard]

you sure you also adjusted the advanced settings? those are very helpful

Did you understand what perf means now or should i try to explain it again?

[north barn]

yup all goods 🙂

i understood what it means now thanks

[simple ore]

then you did not provide it a proper dataset

[ashen terrace]

What MMVS should I download for amd & windows?

[steady mesa]

Hey everyone! My Wokanda keeps messing up with this audio echo there’s a voice in the background that repeats twice.
Anyone else dealing with this or know how to fix it?

[low shard]

This is a General AI Server, AI has many fields, so we can't know your issue with little info

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • E Girl Trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[daring granite]

Does anyone know of any bubbly female voices on Weights ? I'm trying to voice act with RVC and the female voices keep draining the personality away

[low shard]

I think you mean wokada? Be sure you didn't use yt tuts

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • E Girl Trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[low shard]

are you trying to do e girl trolling/catfishing?

[daring granite]

No, voice acting for a skit

pre recorded

[low shard]

You can search rvc ai voice models at:

• https://discord.com/channels/1159260121998827560/1175430844685484042
• In https://discord.com/channels/1159260121998827560/1163592055830880266 , 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
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms

[earnest musk]

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

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[daring granite]

Thank you so much! You're always so helpful and it's really appreciated

[ashen terrace]

your PC GPU:
(AMD Radeon RX 7900 XT)
your operating system:
(Windows 10)
what you want to do?:
Roleplay in Games
what tutorial link are you using:
#1159513888199540817

[pastel orbit]

i have multiple voice models

im trying this one but i don't know why sometimes if you change certain numbers you don't hear the voice it gets muffled and when you play around it just become more audible but not high quality

[ashen terrace]

I downloaded:
MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.16a.zip
and it says 192 (512.0 ms, 24576), but it works for me at 51200 ms.

[pastel orbit]

i still don't understand some other models you use they can work and be audible but slower which is fine since its about the performance and gpu its okay i get that

but this one specifically gets muffled ?

i can't post image here

[low shard]

that is an over year old of original wokada, don't use yt tuts

delete the folder, zip and uninstall from windows app settings

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options GUIDE

• Original Wokada

Not suggested nor maintained, older versions in youtube tuts are even 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 wokada deiteris fork and lmk

[ashen terrace]

But all MMVCServers are over a year old
sorry for ping

[low shard]

wdym? wokada deiteris ofrk latest update is from december 2024, the original wokada you had is from summer 2023 lmao

[ashen terrace]

oh

https://github.com/deiteris/voice-changer

this?

[rapid owl]

hello guys, do u know why i cant add new model on okada

or maybe u guys have okada new ver

[odd shale]

There you have.

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options GUIDE

• Original Wokada

Not suggested nor maintained, older versions in youtube tuts are even 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

[odd shale]

Check either deiteris guide or Vonovox.

[rapid owl]

can i get a new link for amd

i cant find the download button

i want to try vonovox

[flint vapor]

Is just for nvidia

You can use okada deiteris if you want

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/

[hallow thistle]

You have AMD Radeon RX?

Vonovox is only for NVIDIA GPU as of now. Fork W-Okada has both NVIDIA and AMD/Intel variants.

[mental wraith]

Hey, do you know how to get more than three voice cloning artists on the cloud colab ?

thanks a bunch

[hallow thistle]

Using three different RVC voice models at once is not possible on any RVC. RVC can only use one voice model at a time to process. Do you mean like to upload more voice models to a Google Colab RVC notebook or combine/merge voice models into one?

[low shard]

no, you gotta read the guide and get the one from it, not from the github

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[hollow oak]

Can somone help me with Okada voice changer when I’m trying to upload a voice it’s staying on 0%

[waxen terrace]

senpai i want to ask, how to fine tune pretrain model..

my case is training the 4hours indonesian language clean with original rvc pretrain G/D model. this done already..

with this 4hours dataset its really the G and D is already finetuned with indonesian language i train? can i use this G and D model in the next train as custom pretrain? or we need another rules like splitting speaker or other?

i don't really know about fine tuning.

thankyou senpai

[hallow thistle]

I was about to reply to that message, but you deleted it. So, that was not a screenshot, but you implied that the "omegaconf" is flagged as deprecated in Python.

This W-Okada Colab notebook is broken. What is your PC GPU?

No, your computer GPU, not a GPU you selected in Google Colab.

See these guides on how to run W-Okada online. https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-kaggle/ https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-colab/

You're mostly using the "original" version of W-Okada, which is outdated, even the Colab and Kaggle notebooks of it is broken. In the guide I sent, these are the better version than the original one.

[viral mason]

https://cdn.discordapp.com/attachments/1395117266718167091/1399111158526378207/speechmemified_pasteimage.gif

[waxen terrace]

is this named finetuning senpai? or we need more hours of datasets?

[viral mason]

That's not a thing

[hallow thistle]

I'm neither a senpai or sensei. I'm more like a fake ahh sensei.

[viral mason]

Teach me kung fu

[waxen terrace]

[viral mason]

Btw I'm taken dont call me senpai I'm not the guy from fnf

[waxen terrace]

no i mean like, you guys more knowing abt that. so i call senpai haha

[analog obsidian]

4 hours is too small
maybe you could try finetuning 30 hours

[waxen terrace]

ahh i see. but can we continue the G D epoch?

with new datasets but its same character of first

[analog obsidian]

would be better to just use the regular og pretrain, your 4 hour finetune already forgot most of what the pretrain learned before

[viral mason]

That's the old onee

I think

[waxen terrace]

okay so we need a virgin og again, thanks ly

[analog obsidian]

[viral mason]

Oh, it looks like the og one to me but I use the local one so I'm not sure

[hallow thistle]

That doesn't look like the old original W-Okada. Some UIs (slider, selector) just look like that.

[viral mason]

O

[rapid owl]

yeah rx6600 xt

[hallow thistle]

Download and use fork W-Okada DirectML from this guide. https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#download-amd-intel-and-cpu-on-windows

Both W-Okada b2332 on Kaggle and locally look exactly the same.

If input/output devices don't show up on your W-Okada, make sure to give microphone permission on yuor browser or try another browser like mine.

[shrewd forum]

This isnt really help but a question. Is there anything better then the forked version of w-okada or is it the best you guys have right now?

[low shard]

there's nothing better, there's a new windows nvidia only program called Vonovox, which does have some extra features and improvements but not multi platform and as much mature yet
Tho they have similar performance and quality
The main issue is that RVC has been left to rot since 2023 by its original devs, we can't do that much about quality

[shrewd forum]

I see. I heard that rvc v3 was gonna become a thing

Does anyone know what w-okada is coded in? Language wise

[hallow thistle]

Deiteris' fork W-Okada is the best W-Okada after all. There's also Vonovox which is another realtime voice changer but with distinct UI compared to W-Okada one.

[low shard]

Nope, RVC v3 will never happen

[hallow thistle]

Python.

[shrewd forum]

Im planning to just get a team of people to code it if im allowed to 😭

[low shard]

the rvc devs left it to rot, they work on a TTS program called GPT-So-VITS

[shrewd forum]

Since w-okada is open sourced

[low shard]

https://github.com/deiteris/voice-changer

[shrewd forum]

Thats so dumb bro rvc is so good already and they can definetly make it better

[low shard]

you could also work original wokada

[shrewd forum]

Im gonna find alot of people to remake w-okada and make it better

[low shard]

RVC isn't easy, it might also be they hit a wall and weren't able to make it sound much better

[shrewd forum]

Since the forked version isnt enough flr me

[low shard]

even if they "remake wokada", it won't change anything

[shrewd forum]

Possibly

[low shard]

the problem is related to how RVC and the models are trained

wokada is just a realtime voice changer client

[shrewd forum]

I used to make voicemodels but gave up a month in

Its so easy tho

[low shard]

should deffo read the license, but yeah you're able to fork it

[shrewd forum]

Does anybody know one of the devs for w-okada?

[low shard]

making "rvc v3" isn't easy as just making the model, for actually making quality improvements, you have to change core parts of rvc, you could have a talk about this with @simple ore in #🔊│ai-development

[hallow thistle]

Both original and fork W-Okada are mostly coded in Python.

[shrewd forum]

I wanna ask a few questions if I could

[shrewd forum]

No access

[low shard]

If it was that easy to have rvc v3 already, we would have done it already

there's just some devs experimenting

[hallow thistle]

You asked many questions too fast.

[shrewd forum]

And also what happend to the model master shop?

[low shard]

get ai research role

[shrewd forum]

Used to always use it

[shrewd forum]

Alright

[low shard]

Never coming back, paid comms are not condoned in the server, due to legal issues and useless drama

[hallow thistle]

May I please answer at least one of your questions?

[shrewd forum]

Well nick is awnsering them mostly for me so u dont exactly need to but go ahead

[shrewd forum]

I see. Ill miss that lol

Used to always buy from that guy in ur about me

Imcertibtw is my favorite voicemodel creator

Sadly hes quitting

[low shard]

We don't really miss having drama about scams and copyright lol

[low shard]

oh this was just a joke i forgot to take it off 😭

[shrewd forum]

LMFAO

Nah but he made me some insane voicemodels

[hallow thistle]

RVC v3 is unlikely to happen as of now. If you see any RVC fork that claims to be "RVC v3" anywhere on GitHub or Hugging Face, it's mostly RVC v2 in disguise with certain features added as the result.

[low shard]

You should be able to see the rvc development channel now

[shrewd forum]

Btw

Could I get my level roles back?

@fading lodge

This is my old acc

[low shard]

Did you forget your old password?

[shrewd forum]

Yeah..

😭

Got locked out

Email on that acc is deleted + phone number is changed

[low shard]

the only level one is the verified role lol

[hallow thistle]

You have asked 5 questions at once, so that's not a few.

[low shard]

but done

[shrewd forum]

Appreciate it

Bro I remember like

[low shard]

@hallow thistle It's fine for him to ask questions, you're not forced to reply to them btw

[low shard]

I don't remember you much tbh

[shrewd forum]

I applied for mod here and they never accepted/denied my application

Servers dead tbh

[low shard]

The server is actually reviving and better than how it was after summer 2023 tbh

[shrewd forum]

I spoke to the owner of weights a while back and he had barley any info about the original owner bc I was tryna find him

[shrewd forum]

I hope u guys can bring back paid commisions soon 😭

There were some good voicemodel makers on there

[hallow thistle]

The removal of model master shop and the paid request was mostly this. #✨│announcements message

[hallow thistle]

Unlikely.

[rapid owl]

already download but i cant upload the model

[low shard]

Nope, this has been discussed already hundreds of times, and our final decision forever is that we will never bring them back 99.999% lol

[shrewd forum]

Well theres a 0.00001% chance

Better then nothing

😭

[low shard]

They brought alot of drama moderation issues, scammers, weird situations, and other things we don't want to deal with 😅

[shrewd forum]

I hope w okada devs get back into developing this soon

Like this is genuinley the only good rvc client I know

[low shard]

Wok still works on original wokada, but it's mostly only UI changes

Deiteris hasn't been much active since december 2024

dr87 is actually working on Vonovox right now

[low shard]

no matter the client, they can only do so much, the core issue is RVC limitations

like RVC can't laugh, scream, etc

[rapid owl]

help me to upload model pls

[shrewd forum]

Sometimes I can laugh with my voicemodels

[rapid owl]

already load 100% but still blank

[shrewd forum]

Its rare though for it to sound real

Sometimes it sounds great

[low shard]

please elaborate, do you mean in #1175430844685484042 or in a program?

[low shard]

if it was trained on it, they can, but not realistically

[rapid owl]

on program

[shrewd forum]

Im gonna probably spend a hefty ammount of money into finding good developers to make a better forked version of w-okada

[analog obsidian]

it can do some realistic laughs

[low shard]

some

[low shard]

This is a General AI Server, AI has many fields, so we can't know your issue with little info (there are many programs you could be using right now, we dont know which)

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[hallow thistle]

Jokes aside, this has been announced by Vijay himself, so nothing you would expect the paid commission/request to come back to this server. Even if you'd see someone announces their paid commission outside this server, this won't gonna be guaranteed with warranty by mods and admins from this server whatsoever.

[shrewd forum]

I found w-okada back in like 2020 im pretty sure

And been using it since

[shrewd forum]

So im gonna try to make it better since its honestly been the same since I started using it

Ill see if I cab

Can

[low shard]

i think you did a typo, RVC exists since 2023

[low shard]

please elaborate everything I asked

[shrewd forum]

Mabye

I got fond memory of when I started using it

[hallow thistle]

If you have applied for a moderator or helper in this server, but then your application was denied or rejected, it could be either you didn't provide a good answer on why you should become a mod or helper.

[low shard]

again, you could optimize some quality and performance, but still especially on quality you can't do so much unless you actually modify rvc itsself

[rapid owl]

i only have 1 model its so old but works, i want to create or upload another model but it cant. i use win 11 pro, i use rx6600xt

[shrewd forum]

Yeah thats my goal

Alright ima go for now, thanks for the help

[low shard]

you could also check https://docs.aihub.gg/realtime-voice-changer/local/vonovox/ btw, he kinda "remade wokada"

[low shard]

Yw

[low shard]

what tutorial link are you using? can you share a screenshot of the whole program? and what models don't work, what's the exact error?

[hallow thistle]

If you wished to spend your money on a developer who could make a better W-Okada than Deiteris W-Okada, if it becomes a reality, I wanna see how it works and how he or she made it.

[rapid owl]

im not using any tutorial coz i already used for long, here the screenshot of the program, any model cant i use

[low shard]

uninstall vb audio cable, it creates issues on windows, get vac lite https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

set extra to 2.7

can you share the download link of the model you can't upload?

[shrewd forum]

Ah okay this looks new tbh. Ill take a look when im home next week

[rapid owl]

how to set extra 2.7

[low shard]

@rapid owl That model is GPT-So-VITS, not RVC, it's for TTS not STS

also, why would you use a NoMoreSayingCussWords Kid Model?

We don't condone Kids models.

[low shard]

click stop, put the slider of extra to 2.7

[rapid owl]

nah i just random download coz its already 10 maybe different model so i download what i saw

[rapid owl]

so it mean not my vc problem, its models problem

[low shard]

it's not a "problem", it was a TTS model, not STS, 2 different voice model types and programs

also, the model thread has been deleted from the server, we don't allow kids models

[rapid owl]

ahh so i wrong download the models

alright ill try find and download the rvc

[rapid owl]

it works, thanks for the help

[viral mason]

use the slider

[sleek kindle]

So, I wanna get a local LLM but with the vast amount of options im a bit overwhelmed and wanted some advice.
Searching for a model that's a good all in one

My specs are:
RX 9070 XT 16GB Vram
DDR5 CL30 64GB 6000MT/s
AMD 7 9800X3D

Storage shouldn't be an issue. Got plenty

[simple ore]

LM Studio can run LLM using Vulkan on radeon

16Gb is plenty to run 12B models.. and even bigger with smaller quants

or just using llama.cpp vulkan backend

[sleek kindle]

Alright, I'm in the LM Studio store and see a buncha Llama models

[simple ore]

you dont use the store

you use huggingface search

[sleek kindle]

Ah thank you

[simple ore]

like new 'qwen 30b instruct 2507'

IQ2_M quantization

[obsidian rover]

How do I download a voice changer?

[viral mason]

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options GUIDE

• Original Wokada

Not suggested nor maintained, older versions in youtube tuts are even 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

[viral mason]

follow either the first or second guid

as long as you have an nvidia gpu of course

[low shard]

what's your pc gpu? what do you want to do? what's your operating system?

[low shard]

then he has amd integrated graphics lol

better to ask always the basics before helping

[viral mason]

[swift thunder]

help, this appears in my colab : ---------------------------------------------------------------------------
OSError Traceback (most recent call last)
/tmp/ipython-input-3339763087.py in <cell line: 0>()
103 else:
104 backup.unlink(missing_ok=True)
--> 105 model.rename(backup)
106
107 get_ipython().system('rm -r "{LOGS_PATH}"')

/usr/lib/python3.11/pathlib.py in rename(self, target)
1173 Returns the new Path instance pointing to the target path.
1174 """
-> 1175 os.rename(self, target)
1176 return self.class(target)
1177

OSError: [Errno 18] Invalid cross-device link: '/content/Applio/logs/mute_spin' -> '/content/drive/MyDrive/ApplioBackup/mute_spin'

[low shard]

@dull night #🧬│ai-chat message let's keep the convo going here:
Your pc is too Weak for Wokada locally, 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://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/ (but this isn't suggested as it could be unstable)
• Use **cloud **(remote good pc):

About Cloud, there are different services:

• Google Colabs (4 hours daily of free T4 gpu, easy to use, require only a google account) :
  • W-Okada's Deiteris' Fork Voice Changer Google Colab (currently works only on google colab PAID tier)
• 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 (the best and only working one currently for free)

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • E Girl trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program
• the issue

the message above was for someone else about the too weak pc, don't mind it, it's not related to you, there are different methods and programs depending on your needs and hardware

[low shard]

@young blaze it's best you elaborate this

[swift thunder]

Can you tell me how to fix this?

[low shard]

#🧬│ai-chat message
AI needs GPU power, such as a minimum of a gtx 900 serie

[dull night]

@low shard

[dull night]

Thanks

[low shard]

did you check?

You can check your pc gpu on Windows via:

ctrl+shift+esc (task manager) -> Performance tab -> GPU

also which windows version do you have? and what do you want to do? and what's the issue? what tut link did u use?

[low shard]

You're welcome, sorry I just need to follow the rules and make sure the channels are organized, for any further help, please ask here!

[dull night]

For linux, open the terminal and write neofetch

[low shard]

You could technically run it on CPU, but it's not recommended, it will most of the times have high ping, you can try but id rather suggest use cloud
I'm not aware how good your CPU will perform, but if you want you can give it a try, tho I won't garuantee you it will be fine

[dull night]

Sure

Googlw collab?

[low shard]

that only works on the paid tier, you can either try the cpu mode, or pay for google colab, or verify your phone number in kaggle ( a google service) to use it for free

[dull night]

Not problem i will paid, how can i run on google collab?

[low shard]

Are you sure? you can just use kaggle for 30 hours weekly for free

i think 30 hours a week would be good enough, right ?

[dull night]

Yeah, teach me.

[low shard]

I mean, if you say, even tho most people would rather just use the free kaggle one

anyways, you can start to use it by just reading the ai hub docs about the wokada deiteris fork colab https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-colab/, and then ask me for any issue or misunderstanding

[royal cedar]

I am looking for an AI module that animates keyframe animations. from picture 1 to picture 2

[dull night]

i have problem

Installing pre-dependencies...
ERROR: Could not find a version that satisfies the requirement faiss-gpu (from versions: none)
ERROR: No matching distribution found for faiss-gpu
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 261.0/261.0 kB 18.5 MB/s eta 0:00:00
Preparing metadata (pyproject.toml) ... done
Building wheel for pyworld (pyproject.toml) ... done
Installing dependencies from requirements.txt...
ERROR: Ignored the following versions that require a different python version: 1.21.2 Requires-Python >=3.7,<3.11; 1.21.3 Requires-Python >=3.7,<3.11; 1.21.4 Requires-Python >=3.7,<3.11; 1.21.5 Requires-Python >=3.7,<3.11; 1.21.6 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement onnxruntime-gpu==1.13.1 (from versions: 1.15.0, 1.15.1, 1.16.0, 1.16.1, 1.16.2, 1.16.3, 1.17.0, 1.17.1, 1.18.0, 1.18.1, 1.19.0, 1.19.2, 1.20.0, 1.20.1, 1.20.2, 1.21.0, 1.21.1, 1.22.0)
ERROR: No matching distribution found for onnxruntime-gpu==1.13.1
Successfully installed all packages!

<@&1159293140440723499>

[terse halo]

voicechanger stopped working......

[dull night]

alternative?

[terse halo]

no

I only have this one

I tried to use it now and nothing happens

[low shard]

@terse halo that's the local wokada deiteris fork, they are using wokada deiteris fork colab

[low shard]

could you try ignoring that error and going on with the rest of colab?

[low shard]

also, no need to ping mods, Helpers are the ones who help

[dull night]

Doesnt working

[dull night]

okay

[terse halo]

but why doesn't it work anymore?

[simple ore]

hm... perhaps a problem with the latest colab update and you using 3.2.9 version

[terse halo]

wait, deiteris wokada doesn't exist anymore?

[dull night]

[Errno 2] No such file or directory: '/content/voice-changer/server'
/content

---

ModuleNotFoundError Traceback (most recent call last)

/tmp/ipython-input-4113096174.py in <cell line: 0>()
22 get_ipython().run_line_magic('cd', '/content/voice-changer/server')
23
---> 24 from pyngrok import conf, ngrok
25 MyConfig = conf.PyngrokConfig()
26 MyConfig.auth_token = Token

ModuleNotFoundError: No module named 'pyngrok'

---

NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.

To view examples of installing some common dependencies, click the
"Open Examples" button below.

[low shard]

the docs have been updated, along with some files, the new link is https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/

[low shard]

I will check myself rq

[dull night]

there's no alternative?

[terse halo]

Will I have to download the new version for it to work?

[dull night]

okay

[low shard]

what are you talking about? are you having issues? the latest wokada deiteris fork is b2332, from december 2024

[terse halo]

I'm saying that the voicechanger is no longer working

I tried to use it now and nothing happened

[dull night]

In short, can I use this software anymore?

[terse halo]

when I click start, and speak, the AI doesn't say anything

[naive bramble]

does anyone know how to use a rvc model for text to speech not real time to make a audio file in high quality dont really know how to do it

[low shard]

Yeah, by paying for HuggingFace pro , you can't expect unlimited free gpu usage, it's expensive for huggingface

[low shard]

There are different Text To Speech (TTS) AIs:

• GPT So Vits: Great Few Shots (needs a lil training) TTS, its only limited to: english, chinese, Cantonese, japanese & korean, if you wanna check gpt so vits instead, read https://docs.aihub.gg/tts/gpt-sovits/
• 11labs: Easy way to do TTS is https://elevenlabs.io/, its a mostly premium easy way for good quality TTS
• FishSpeech: FishSpeech is a 0 shot (no explicit training needed) TTS, if you got a good pc you can use it locally else use their site

You can check other TTS in our tts index

With RVC Models:

RVC is natively for Speech To Speech, but forks such as Applio have built in tts (using Microsoft Edge TTS to make a tts audio, i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• You could try another tts from our tts index and use the output as an input in rvc

[terse halo]

Do you know why the voicechanger might not be working?

[low shard]

what's your pc gpu and operating system? what are you trying to do?

[naive bramble]

is there any way to convert a rvc model to tts

or

[low shard]

set extra to 2.7
show also your discord (or whatever program you're going to use this in) settings

[low shard]

did you maybe miss to set your Ngrok or Horizon token? it works fine

[terse halo]

I tried to put it on 2.7 and it still doesn't work

[low shard]

set input to line 1 on all other programs you want to use wokada deiteris fork in

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Advanced Settings -> Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Advanced Settings -> Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Advanced Settings -> Crossfade Lenght: Controls how smoothly the AI stitches different processed parts "chunks" of your voice back together. 0.1 for fastest voice, 0.15 for improved quality but increases delay by ~50 ms
• Reduce the delay on Windows via the Wasapi / Asio Guide

[low shard]

Nope, those are 2 different ai projects and architectures, they work in 2 different ways

[terse halo]

I put it in but it still doesn't work...

[naive bramble]

is there anyway to upload your own models to eleven labs etc?

[dull night]

I am following this document;
https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/

its correct?

[low shard]

Macs aren't great for AI, and RVC doesn't have great support either, but you can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides, probably won't be able to train, make models):
  • 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 limited time):
  • Weights.com: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Applio Colab: max 4 hours, not granted, of GPU
  • RVC-AI-Cover-Maker-U Colab & Kaggle: Automatically separates the vocals and instrumentals, converts the voice and mix all together back
  • Ilaria RVC Zero: fastest and simplest that you can get for free

Easiest possible (automatically separates vocals & instrumentals) : weights.com & rvc-ai-cover-maker-ui colab/kaggle
Easiest cloud: Ilaria rvc zero
Easiest Local: Applio

[terse halo]

I set it to listen to the AI speaking, but when I speak, the AI doesn't reproduce, what could it be?

[low shard]

nope that's not correct, that's for local wokada deiteris fork, you need to follow the wokada deiteris fork colab guide: https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable, since your pc isn't good enough to run it locally

[dull night]

dude i installed yum package manager for this!

[low shard]

you sure you did this step?

can you show me your recording and playback tabs?

[terse halo]

the program was working before, it just stopped working now

[dull night]

should I download all the files in both cpu and gpu section or only cpu?

[terse halo]

Could it be that one of the Win 11 updates caused VoiceChanger to stop working?

[low shard]

yep, you missed that step
in both recording and playback tab: you need to click your usual devices, then set as default, then right click them and set as default communications devices

[terse halo]

ty

[low shard]

I think you're confusing things up

Local: runs on your hardware
Cloud: using remote good pc services, such as Google Colab and Kaggle, suggested for low-end hardware users

You got some servers cpu, you could try locally but I can't guarantee you it will be fine
Or you could use cloud (kaggle free 30 hours weekly with phone verification, or google colab but only in paid tier)

What exactly you want to do?

[low shard]

is it all solved now?

[terse halo]

yes

[low shard]

yw

https://tenor.com/view/hatsune-miku-triple-baka-braincells-gif-17994396760098916322

[naive bramble]

is there anyway i can download the required voice files for tts can anyone guide me where do i find that

[dull night]

how to kaggle or colab?

working

[low shard]

wdym exactly? are you trying to convert RVC models to TTS again?

it would be better you elaborate, there's different TTS programs

[naive bramble]

no sorry i just want to download seperate stuff for tts

thats what i meant

forget rvc

[low shard]

I will give you both guides for both!:

• Google Colabs (4 hours daily of free T4 gpu, easy to use, require only a google account) :
  • W-Okada's Deiteris' Fork Voice Changer Google Colab (currently works only on google colab PAID tier)
• 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 (the best and only working one currently for free)

[low shard]

what's your pc gpu and operating system? which tts language do you need?

[naive bramble]

5070ti and win 11 pro and languages im not familiar with sorta new

but if its about the voice english

[dull night]

thats working, right?

[low shard]

do you also need voice cloning? or just need the most basic easy TTS?

[low shard]

yup, both of those are working

[naive bramble]

most basic as in>

?

i dont think i will be voice cloning

generally i will prob just use the voice for dialogues

[low shard]

like easiest with premade voices only lol

[naive bramble]

yeah premade voices

i dont really think i will be training anything anytime soon

[dull night]

I can verify with turkhis number?

[low shard]

https://docs.aihub.gg/tts/tts-tools/#edge-tts
edge tts:

• multilingual
• highquality
• with premade voices,
• no voice clone
• not emotional
• cloud only (there aren't really liimits tho it's basically microsoft edge tts reader)

Do you think this would fit you?

[low shard]

I think so, it's better to try

[naive bramble]

this is in ms edge or i can use other browsers as well

[dull night]

verification code dont comed

😅

[low shard]

You can either use it from:

• the browser
• google colab
• huggingface space
• rvc forks (with rvc models):
  • applio locally
  • applio colab
  • ilaria rvc zero (currently half broken)

[low shard]

could you try again? you could also try contacting kaggle support https://www.kaggle.com/contact or using colab, or another number

[naive bramble]

i ran the thing followed the process renamed to html opened it just need to know which models i need to download for tts like what type of model like rvc is for the other thing what do i need to get for this and how do i use it on the browser

also do i have to do something with audacity recorder

or

[low shard]

if you're asking how to use it, please read https://docs.aihub.gg/tts/tts-tools/#edge-tts
also, the premade voices already come in with edge tts, you don't manually downlooad the voices like in rvc

[dull night]

i used virtual german number, thats work!

[low shard]

great!

[dull night]

how to install Virtual Audio Cable on linux?

endavour linux (pacman package manager) and evolinuxos (apt package manager)

@low shard

[low shard]

I just updated the docs, for some reason the linux way didn't show on the cloud ways, can you try refreshing https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-kaggle/#virtual-audio-cable now?

[dull night]

okay

[dull night]

Help

Please

[low shard]

Be sure you ran the Arch based linux command btw

[low shard]

gpu: one of the T4 gpus
extra: 2.7
chunk: 100
f0: rmvpe without onnx

and you can add any voice model you want

[dull night]

I'm running dual boot my arch based endavour system, now I'm on ubuntu noble numbat based evolinux and installed accordingly. no model how to add it

[low shard]

You can search rvc ai voice models at:

• https://discord.com/channels/1159260121998827560/1175430844685484042
• In https://discord.com/channels/1159260121998827560/1163592055830880266 , 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
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms in the server

[earnest musk]

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

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[dull night]

My English is so bad! Can you throw me the gawr gura model?

[low shard]

are you looking to do roleplay or e girl trolling/catfishing?

[dull night]

I have a friend who is addicted to anime, he likes to roleplay

[low shard]

I see, https://huggingface.co/dacoolkid44/VTuber-RVC/tree/main/Gawr Gura (Talking)

[dull night]

Thanks

[dull night]

How do I add this?

[low shard]

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#adding-models read only this part of this guide

[dull night]

Thanks!

I will try

now

[low shard]

is it working?

[dull night]

not really, it just makes my voice thinner.

[low shard]

can you show a screenshot of ur wokada and other program settings?

[gray urchin]

uhh guys whats wrong?

oh i cant send prints

its like im using w-okada RVC and its like 30k ping even on client

[wet orbit]

Anyone know why my advanced settings are like greyed out?

It wont let me change any of them for some reason on the V2332 client

[knotty moth]

stop the voice changer first

[wet orbit]

Bingo, Thank you @knotty moth

[knotty moth]

there are many gura models you can search in https://discord.com/channels/1159260121998827560/1175430844685484042 and weights.com

then try them to find which may sound the best for you

[dusty rampart]

Wah

[humble hornet]

when i launch w-okada my mic goes like laggy almost stuttery

[oak edge]

hey with rvc here are 2 models I'm training with same 40mins+ data set, (one I'm using default pre train and blue one custom Titan 40k pretrain), and this is loss/g/total with 0.987 smoothed, is blue one better in terms of voice swapping than grey?

[knotty moth]

make sure you have a capable gpu with correct chunk & extra settings

you can refer to this guide for less audio latency

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay-windows-only

[tawdry shell]

I have the json of a voice model but not the index file(which I need for applio) any way I can get the index from an existing model already?

[signal mauve]

how do I avoid the robotic voice effect in okada

What settings should I use to make the voice sound more natural

[potent lichen]

I know this is server is mostly related to audio but is there any AI that can help remove text from a video?

[hallow thistle]

Which W-Okada version are you using and what is your PC GPU?

[signal mauve]

idk what is my verison and my gpu 3050 rtx

[hallow thistle]

Are you using this W-Okada?https://cdn.discordapp.com/attachments/1159290139609137264/1400869547178852412/image.png

[signal mauve]

yes

[hallow thistle]

The b2332 and v.1.5.3.18a are different W-Okada versions. Also, here are your chunk and extra settings if you use fork W-Okada (b2332). https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#known-working-settings-for-chunk-and-extra

[grave swift]

mine sounds so messed up

[dull night]

Please wait 1 hour

[paper widget]

how can i clone and TTS for non english langs locally?

[indigo iris]

heya was wondering how to get the model to speak through the text to speach on windows 11

[humble hornet]

4080 is what im using and for input i cant see cable input only output

[knotty moth]

tell me for image perms to show the screenshot

also what kind of game were you playing?

[covert thicket]

i need HELP!
whats the best voice model , that sounds exactly like a girl
i wanna feel like a girl

[simple ore]

trying to catfish, eh?

[covert thicket]

No i will not wthhhh

i identify as a girl

[low shard]

This is a General AI Server, AI has many fields, so we can't know your issue with little info

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do?:
  • AI Covers
  • TTS
  • Roleplay in VC
  • Roleplay in Games
• what tutorial link are you using / a screenshot of the program

[low shard]

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do? There isn't a program that does everything, there's a program for each thing:
  • AI Covers
  • TTS
  • Roleplay in VC
  • Roleplay in Games
  • etc
• what tutorial link are you using / a screenshot of the program

[low shard]

i'm guessing you got the model off weights.com:
in rvc context:

• pth files: contain the voice
• added index files: contain the accent
• metadata.json 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

so no, you can't get the index, you can use the model without the index

[low shard]

what's your operating system? what tutorial link did u use?

[low shard]

This is a General AI Server, AI has many fields, so we can't know your issue with little info

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do? There isn't a program that does everything, there's a program for each thing:
  • AI Covers
  • Train RVC Models
  • TTS
  • Roleplay in VC
  • Roleplay in Games
  • etc
• what tutorial link are you using / a screenshot of the program

[low shard]

alright

[low shard]

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• which language do you want to do?

[low shard]

wdym? can you elaborate on what you want to do and what's your pc gpu please?

[low shard]

ohh you're trans?

You can search rvc ai voice models at:

• https://discord.com/channels/1159260121998827560/1175430844685484042
• In https://discord.com/channels/1159260121998827560/1163592055830880266 , 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
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms in the server

[earnest musk]

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

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[low shard]

set f0 to rmvpe without onnx,
play with the pitch
be sure the browser has microphone perms
i'm not so sure how the input output would work for linux, but the input should be ur microphone, and the output the VAC, while the other program (like discord)'s input should be the VAC

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Advanced Settings -> Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Advanced Settings -> Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Advanced Settings -> Crossfade Lenght: Controls how smoothly the AI stitches different processed parts "chunks" of your voice back together. 0.1 for fastest voice, 0.15 for improved quality but increases delay by ~50 ms

[dull night]

Still, it doesn't work on my voice for some reason, it just makes it a little thinner.

[low shard]

try everything i said, try other models too, be sure the vac is working

[dull night]

vac?

[low shard]

Virtual Audio Cable, you got portaudio right?

[dull night]

I installed it but it doesn't show up as an app.

[covert thicket]

ye but i identify as a girl

she/her

[low shard]

you can try searching in the sites or ways i told you, just know rvc models got limitations and can't super realistically perfectly all non speech sounds

[low shard]

show all options you see when clicking output

[covert thicket]

idk anything about it >.<

[low shard]

wdym? do you need also a realtime ai voice changer for calls or just the rvc models?

[covert thicket]

yes a realtime one so i can use it in discord 0w0

[low shard]

what's your pc gpu? operating system?

[low shard]

@analog obsidian @pastel oak you got any idea on how the vac works for linux?

[covert thicket]

gtx 1650 and windows 10

[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 WebUI with the best general support for many platforms. GUIDE

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options GUIDE

• Original Wokada

Not suggested nor maintained, older versions in youtube tuts are even 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]

you can either use wokada deiteris fork or vonovox

id suggest reading the pros and cons of both of them

[pastel oak]

Use portaudio, its on the guide

[dull night]

already installed

[pastel oak]

Check the GitHub page in issues section if someone asked about it

Think someone did before

[brittle wing]

@simple ore well thing is i had this weird shee with 7900xtx that whenever i add any type of index my previous cpu would bug the hell out

and during heavy games id like the be able to use it normally aswell so thats a reason to want to use NPU

[simple ore]

still download that DML fork for realtime

see what canbe selected in the device option

[low shard]

#🧬│ai-chat message adding this for context

can you share a full screenshot of the program you're using?

[simple ore]

you can also run applio on it with a little hack

[simple ore]

there you have it, pick one or the other and see how it goes

[brittle wing]

wich one will i need to make use of the NPU tho or is it an automated process?

[low shard]

yeah you're already using the wokada deiteris fork b2332,
set input: microphone
output: line 1
Monitor: headphones optionally to hear urself
extra: 2.7
chunk: a bit higher than perf

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Advanced Settings -> Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Advanced Settings -> Crossfade Lenght: Controls how smoothly the AI stitches different processed parts "chunks" of your voice back together. 0.1 for fastest voice, 0.15 for improved quality but increases delay by ~50 ms
• Reduce the delay on Windows via the Wasapi / Asio Guide

[low shard]

the intel graphics should utilize the NPU right?

[simple ore]

supposedly 😛

[brittle wing]

ok lemme get a model rl quik and see

[brittle wing]

did these settings btw tyty

[low shard]

this is interesting

[simple ore]

I mean Zluda been there for a year

[low shard]

I asked you smt about it in #💬│staff-chat , if you want

[low shard]

yw

[low shard]

lemme know how it goes, it woulld be interesting if it can use the NPU

[brittle wing]

w8 chunck higher or lower then 512 or just mess around and find out type sh

[low shard]

just make sure the chunk is a bit higher than the perf value at the top left

that's the rule of thumb for every device lol

[brittle wing]

ok so on intel R graphics selected it does not seem to be using NPU

in task manager

[low shard]

so, it seems to be using only the integrated graphics, right?

[brittle wing]

yup

[low shard]

well, i don't think there's much you can do then in this case, NPUs are severly less supported in AI compared to GPUs, it's rare someone has one

[brittle wing]

im just tryna indicate what performance NPU would have over lets say my intergrated graphics

since its built for ai

[low shard]

would be better than integrated graphics, would be worse than your GPU probably

maybe, check for any intel ultra drivers (https://www.intel.com/content/www/us/en/support/detect.html) , amd gpu drivers and windows updates
Then, restart your pc, and show a screenrecording of you trying to utilize the NPU with task manager opened

[molten pilot]

Hello i have a question
Where i can find french (vf) RVC model (voice model)

[low shard]

You can search rvc ai voice models at:

• https://discord.com/channels/1159260121998827560/1175430844685484042
• In https://discord.com/channels/1159260121998827560/1163592055830880266 , 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
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms in the server

[earnest musk]

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

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[molten pilot]

I want french one

[low shard]

there are thousands of rvc models, you can search them yourself

[molten pilot]

Which site si u recommende

Omg my english so bad

[low shard]

you can use any, the quality doesn't depend on the site
however the ones with the most models might be https://weights.com

[molten pilot]

Ty

Ty

I will search

Hmm

I installe the app?

On my phone

?

[low shard]

you can use either the weights app or site

[molten pilot]

Okk

[ember lily]

https://huggingface.co/spaces/TheStinger/Ilaria_RVC why ilaria rvc doesnt work?

[low shard]

it's half broken, read #📰│dev-updates message, it's unlikely it will comeback anytime soon, it's better you use an alternative

[viral mason]

https://tenor.com/view/star-wars-ben-quadinaros-gif-11772983869383392013

[ember lily]

Thank you so much. There is an alternative like google colab or another things related to hugging face?

[low shard]

yeah they are all linked in the message
if you tell me your pc gpu, operating system and what you want to do, I can help you personally btw

[ember lily]

Thank you so much. I have a mac air m1 8gb, I basically need only to convert an acapella into ai vocals

[crisp kettle]

just a quick question, if I try multiple kaggle account, I could just switch between them to have more gpu usage right?

[low shard]

For Inference (use models) Mac (which doesn't have great support), You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides, probably won't be able to train, make models):
  • 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 limited time):
  • Weights.com: Partnered with AI Hub, lets u do them easily but u may be in a queue
  • Applio Colab: max 4 hours, not granted, of GPU
  • RVC-AI-Cover-Maker-U Colab & Kaggle: 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 colab/kaggle
Easiest cloud: applio colab
Easiest Local: Applio

[low shard]

You can technically do that (multiple accs, multiple emails, multiple numbers), though it's against their ToS and might get detected and they could possibly take action on this

[ember lily]

Thank you very very much🙏🙏

[low shard]

You're welcome, let me know for any issues

[crisp kettle]

how would they know? through my IP? Anyway to reduce the risk?

even if I use them frequently, two accounts are enough per week

[low shard]

through my IP?
possibly, not sure lol

just know that by doing that you take your own personal risk, we can't do anything nor condone or get responsible for this, we don't work for kaggle

[simple ore]

@low shard "This is Intel's first desktop processor with an NPU, but it isn't the latest NPU 4 from Intel that you find on Core Ultra 200V Lunar Lake mobile processors, but rather the older NPU 3 unit from Core Ultra 100 series Meteor Lake, which can only do 13 TOPS. It hence misses out on Microsoft Copilot+ native acceleration"

new cpu, shitty npu

no wonder Intel is having troubles

285k is algo meh

[gusty bison]

hey i just installed the software but the voice is getting way tooooooo delay and it's like choppy not really clear

[low shard]

This is a General AI Server, AI has many fields, so we can't know your issue with little info

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do? There isn't a program that does everything, there's a program for each thing:
  • AI Covers
  • Train RVC Models
  • TTS
  • Roleplay in VC
  • Roleplay in Games
  • etc
• what tutorial link are you using / a screenshot of the program

[next pecan]

hi the voice changer is working but im getting no sound

from it

there's no audio coming from the voice changer

[low shard]

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do? There isn't a program that does everything, there's a program for each thing:
  • AI Covers
  • Train RVC Models
  • TTS
  • E Girl trolling/catfishing
  • Roleplay in VC
  • Roleplay in Games
  • etc
• what tutorial link are you using / a screenshot of the program

There's different programs and versions, we can't know exactly the issue with so little info

[next pecan]

gtx 1660 super
wi 10
uh im just testing it, i wanna sing in gokus voice
how do i check the version

its the fork one

[low shard]

i wanna sing in gokus voice
are you talking about singing in discord vc with friends, or like use it for a pre-recorded audio? because you might have been using the wrong program

[next pecan]

the fork one

not the pre-recorded

[low shard]

that's a different question, you might be using the wrong program for your use case

also, "fork" just means "modified version" in general I.T. terms
you'd need to share a link of the guide and a screenshot of the entire program

!give-media-perms 1h @next pecan

could you please carefully explain also what you want to do exactly? RVC stands for Retrieval-based-Voice-Conversion, not realtime voice changer

[next pecan]

uh i wanna sing in discord voice chat

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/

[low shard]

share a screenshot of your entire wokada deiteris fork (advanced settings included), and discord settings

[low shard]

can you please share the entire thing? without cropping

[next pecan]

ok

this?

[low shard]

i meant also above scrolling, at the top there's the type version, to be sure you'e using the right one

[next pecan]

b2309 nvidia cuda

?

[low shard]

that's not the right one, that's a version off Aug 25, 2024

you somehow downloaded an older version of wokada deiteris fork

you sure that's the one you got? could you please show it?

@next pecan if that's the version you got, you should delete it

read https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#download-nvidia-on-windows carefully, be sure you get b2332 from december 2024

[simple ore]

1660 super

might as well throw it out of the window

[low shard]

i mean it works with wokada deiteris fork decently unless in an intensive game

[golden walrus]

guys, do you know if pretrain affect the voice tone ?

and pitch

[simple ore]

the voice should not change from the dataset, the pitch may improve while inferring audios outside of the datasets's pitch range

[golden walrus]

i still have no idea why 1 sound nice but can't spell, and 1 sound like a kid

[simple ore]

but it is better to have some singing if you want the model to sing

[golden walrus]

my brain is on fire rn

[golden walrus]

but i will keep this advice hehehehe

[analog obsidian]

mispronunciations? oh thats contentvec's fault

[golden walrus]

let me train this one again

but yes, the old one sound nice but i trained it with cvec

[fleet cedar]

what the hell gpu is that 😭

[undone orbit]

Hi guys, how can I post a screenshot in this chat? I'm struggling with running the Eddycrack864/
UVR5-UI https://i.ibb.co/PsMPGfLz/2025-08-02-22-47-54.png https://i.ibb.co/33tVY73/2025-08-02-22-51-33.png

[dull night]

I don't understand

[low shard]

@dull night @fleet cedar that's the Kaggle T4 GPU

[low shard]

!give-media-perms 1h @undone orbit

now you can send images lol

can you explain please

• your pc gpu
• what's the issue?

-# i'm guessing you're on windows 11

[undone orbit]

• Windows 11 Pro, Intel i7-14700K, RTX 5060i 16GB
  -Downloaded pre-complied version of UVR UVR5-UI-v1.8.4.zip
  -Installed it
  -It opens as per screenshot, however no matter what model I select, I get this message that model has to be downloaded and nothing happens.

Anyway, I just realized that models can be downloaded in settings so I did that. Shut it down to restart it and now when trying to run it again I get this message

[golden walrus]

you wait

it shows nothing in the cmd

just wait

just expand the batch seperation and you can see the progress

also, you don't need to download audio again when finish, it's in outputs file

[undone orbit]

I can't even open it anymore 🙁

[golden walrus]

run the install one then re run, i met this issue just a while ago hahahaha

I assume you download the compiled version

Like me

[low shard]

@viscid moss are you aware of this issue?

[low shard]

yes, they said that lol

[golden walrus]

Me didn't see

[nimble kelp]

when i use the voice changer i hear myself when i talk

how do i stop that

[undone orbit]

I instantly get errors 🙁 really sorry for the spam guy, but this is driving me nuts for few days now

@golden walrus what should we put in these fields? Also, do I need to intall FFmpeg for UVR to work?

[low shard]

Please Elaborate:

• your PC GPU (Such as Nvidia RTX 3060, AMD RX 9070 XT..)
• your operating system (Like Windows 10, Windows 11, Linux, MacOS 26..)
• what you want to do? There isn't a program that does everything, there's a program for each thing:
  • AI Covers
  • Train RVC Models
  • TTS
  • Roleplay in VC
  • Roleplay in Games
  • etc
• what tutorial link are you using
• a screenshot of the program

[golden walrus]

no, you just need to put audio in and press seperate

[undone orbit]

Making progress, just to double check (because I don't see options to select GPU utilization like in standard UVR that you download from google) this version definitely uses my GPU for separation right? 🙏

[golden walrus]

i have no idea. I have it run on gpu

[broken eagle]

can someone help pls

[simple ore]

it is a free GPU on Colab/Kaggle, pretty old

[viscid moss]

@undone orbit Haai are u still having issues?

[undone orbit]

Hi @viscid moss thanks for getting back to me. Stuck on the queue for good 15 minutes now and don't think GPU is being utilized. Unfortunately, cannot post screenshots anymore 🙁

[viscid moss]

!give-media-perms 2h @undone orbit

[viscid moss]

Ok, so now u can post images

Let's see... u have a 5060 rigth?

[undone orbit]

yes, 5060ti. Please let me know if you would like to see something specific

[viscid moss]

yah, pls go to UVR5-UI/info and run status-checker.bat

[viscid moss]

seems like u don't have ffmpeg installed, u need that

[simple ore]

the usual issue

[viscid moss]

run UVR5-UI/info ffmpeg-installer.bat

or u can install it manually, as u wish

[simple ore]

can grab 2 exes from https://huggingface.co/IAHispano/Applio/tree/main/Resources

and drop them into uvr folder

[undone orbit]

done, now I get this when trying to run it. I have to restart the PC and then it will work (tried before)

[viscid moss]

I'll add those to the precompiled version next time :3

[viscid moss]

How many CMDs u have open? I think u re trying to run the app twice

[simple ore]

kill python.exe in task manager

[undone orbit]

looks like we are cooking on gas

[viscid moss]

Nice!