✨│ai-help

[low shard]

videos can't be updated like written guidesd

[stiff valley]

bro whenever i do something myself i mess up absolutely everything

[low shard]

you're missing out on alot of better quality and performance

[stiff valley]

how can i do it correctly?

[low shard]

uninstall everything you got off that youtube tutorial

then

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[low shard]

read up the 1st link

wokada deiteris fork >>>> original wokada shortly explained

it's more optimized for better performance and has ways to get more stable models

which original wokada currently dont have

[stiff valley]

bro what to press after opening the link?

[low shard]

you gotta read it

you have to read it to download and utilize the program

if you don't or skip parts, it might give errors so please be careful

[stiff valley]

well this is the first step and i already dont know where to download it

[icy ember]

I still got some issues, the delay and some robotic voices when playing

[low shard]

you got the rtx 40 serie, that part is for the rtx 50 serie, download the normal rtx one

[stiff valley]

BRO WHAT

[low shard]

set f0 rmvpe without onnx

[low shard]

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

[icy ember]

🫡

[stiff valley]

does this atake olong?

oh nvm its done

what now

no matter what voice i put in it just sounds like a robot

[low shard]

set extra to 2.7, chunk to 140, f0 to rmvpe without onx

did u install vac lite, right?

if not, check it in the virtual audio cable guide part

[stiff valley]

i did

do i use it somehow?

[low shard]

set output to line 1 in wokada

then, in the other program u want:
input: line 1
output: headphones

[low shard]

be sure to get a good voice model, not all are good

also, u can optionally get better quality and stability by turning force fp 32 mode on in advanced settings but at the cost of some more delay

[stiff valley]

what if i do it in game?

do i then do it through windows settings?

[low shard]

same shit in the game settings

not the windows settings, leave them

[low shard]

no, just via the game settings

[stiff valley]

hmm in game there is like a 10 second delay

also is there a way to have it windowed, and not opened in browser

[low shard]

close every other program in background, put game graphics to the lowest ever possible

[low shard]

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

[stiff valley]

aaa okay

thanks g i think i understand everything now

[low shard]

are u interested in following a bit more complex process with server mode to get lower delay?

[odd isle]

the lower the perf the better right?

[fluid birch]

Hey everyone! I am looking for an open-source software that will allow me to use voice models to edit recordings, voice-to-voice. Anyone know where I can find something like this?

[icy tinsel]

Hello there, excuse me, how can i run w-okada (that is running locally on my pc) over web? I want to run realtime conversion with my phone, with separate mic and speakers, for a cosplay. Is there any realtive rvc that can work with phone?

[uneven harbor]

Hey everyone, I just launched LifeAI – a simple AI chatbot that runs directly in the browser:
https://lifeai.vercel.app

It currently supports Google login and uses a custom backend to handle prompt/response logic via API.

I’d love your thoughts on a few things:

1. Latency management – How do you handle real-time feedback or streaming response UX for AI responses without overloading the frontend?
2. Token limits / context awareness – Any tips on managing longer conversations or summarizing history smartly?
3. Security for public AI endpoints – How to best protect from abuse if user sessions are lightweight?
4. Prompt engineering strategies – Are you using roles/personas/templates that work best for general users?
5. Monetization – What’s the best approach for small chat-based AI tools? (Ads, freemium, credits?)

Open to ideas and collaboration. Let me know what you'd improve if this were your project.

Thanks in advance!
support mee: https://trakteer.id/lifevander🥰😇

[low shard]

the lower the perf, the less delay yes

[low shard]

what's ur pc gpu and do u need it for pre-recorded audios or realtime for calls

[distant turtle]

-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

[flint anvil]

ok so i have two computers, one with a really good gpu - how do i make okada's rvc client work as a server and have one computer do the actual conversion so i can max out the settings and have another computer do whatever? i know its a feature in the software i just dunno how to activate or use it and docs are scant on it

[cold furnace]

Hey all, so i got a pretty high-level question, not looking into implementation details here, just maybe platform or tech?

i guess im not the only one seeing the "Peter and Stewie tech reels" popping up everywhere lately, some of them sound really clean and not just sound like them, also speaks like them tone and pronounciation wise.

does anyone know how they do that?
i tried using RVC and voice cloning but i managed to make it sound like them, not speak like the characters really.

thanks in advance!

[copper terrace]

can I get the software link that is used to redirect the voice from the voice changer to other output

[simple ore]

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

[brittle wing]

anyone has 2070

?

i need some settings for it

[vale osprey]

-colab

[patent trellis]

📒 Google Colab Notebooks

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

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

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

by Jarredou & Makidanye
Google Colab

[brittle wing]

any idea why it could not send audio to line 1 but i can hear it using the monitor i tried reinstalling the line 1 still the same thing idk what to try more

[flint anvil]

can anyone help?

[brittle wing]

Explanation on GitHub can be found here: https://github.com/deiteris/voice-changer/issues/180#issuecomment-2359166278

In short:
You create a file named .env on the same folder where MMVCServerSIO.exe is located. Open it up with a notepad, copy paste the settings from the GitHub link.
After that, you create another file with the file extension ending .bat, open it up with a notepad, copy paste what is needed in there again from the GitHub link.

Now run the bat file. After it starts, you should be able to open the link. For example, if you specified HOST=192.168.0.1 and ALLOWED_ORIGINS='["https://192.168.0.1:18888"]'), you should be able to open https://192.168.0.1:18888 in your browser and use the voice changer UI from other machines in your local network.```

[flint anvil]

ok so if i open up the ui from pc 2 (if pc1 is running the rvc server), i set inputs/outputs to the mic/virtual cable from pc1, the audio should go from pc2 => pc1 => pc2 again? just making sure i understand
also should i still have the setting as "client" on the ui itself so i can put in the mic & virtual cable?

[slim schooner]

why not just use the pc with the good gpu for everything? or is trying to use both pc's more of a personal preference?

[brittle wing]

so it works on original okada but doesn't on deiteris

idk what to do atp

[simple ore]

client is your weak PC, browser takes audio from the mic and sends it to the remote server for conversion, it gets it back and send it to the virtual cable on the weak pc

[simple ore]

in case you still have questions, OpenVoice v2 is trash

[shadow sparrow]

anyone understand this? started running the applio colab with the gradio method and it keeps saying to put this file into the location but WHERE is the location cause im not running this on a program, its just running through a colab

[simple ore]

that should not happen

[shadow sparrow]

it was working perfectly fine an hour ago and now just keeps saying to download this file and put it in this even though how tf do i do that if im running it through a google colab 😭

[simple ore]

lemme check, one min

[simple ore]

nope, everything is fine

[shadow sparrow]

randomly fixed itself now, weird

[pulsar kernel]

can someone help me with training an ai

im trying to train an ai but when i click process data the output is blank

[azure patio]

is there anyway to get the .pth file if the runtime is dissconected?

like the one after 300 epochs for example

[pulsar kernel]

@azure patio it should be in your drive folder if you had a low enough save frequency assuming ur using google collab

[icy ember]

still not good enough 😦 it spikes so bad

[low shard]

set chunk a bit higher

you can’t expect 0 delay

u can try https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#reduce-more-delay

[icy ember]

Thanks

[low shard]

yw and lmk

[icy ember]

Well i don´t really know where to go, since its been a while, i think a pc reset might solve the problem. At least that is what i hope for.

I cant tell what causes the performance issues, sometimes is super fine, sometimes it looks like this.

[low shard]

are u using it in a game

elaborate when the issue happens

[icy ember]

It happens when im ingame yes, but i dort know what exactly causes the problem.

Richtig now the server is empty and i was next to 1 Person and i had a delay of over 3 seconds and more

Then, when im with a Group of people i dort seem to have a problem

[low shard]

that’s the thing, your gpu is doing 2 extremely heavy tasks at the same time

set the game graphics to the literally most lowest ever till you can even count the pixels basically

also cap fps to 30 or 60

[icy ember]

I got a 60 cap atm

[low shard]

are the graphics the lowest ever possible?

[icy ember]

Will try that, bit the i gwt the Probleme that the game laggs. If the settings are too low lol

[low shard]

what game are u possibly playing

try googling the best setting sfor like potato pcs for that game if u want

[valid spindle]

Is there a place for recommended settings for the Okada vc?

[low shard]

what’s your pc gpu?

[icy ember]

ok

[valid spindle]

4060

[low shard]

be sure to never use video tutorials

if you used a video tutorial for wokada, you wasted time since they ar eold

[valid spindle]

❤️ ^^

[low shard]

alr, did you already follow the written guide of wokada deiteris fork?

[valid spindle]

😅 I followed a tutorial on youtube on how to download the software then I did a bit of customizing by my self

I was looking at the github to learn about the different settings and downloaded some voices fromn the discord

[low shard]

uninstall all you got from it, you 100% got old original wokada and vb audio cable

And both of those cause issues

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[low shard]

read up the wokada deiteris fork guide, the 1st link

[azure patio]

Theoretically yeah but i dont have a folder at all in drive for some reason

[valid spindle]

Alrighty will do <#

[low shard]

no, an intel cpu won’t be an issue

it will run on ur rtx 4060

[hot lagoon]

What is best for vocals right now

[valid spindle]

Hey so I download what I was told to download on the guid and I dont see the option to run the program

I downloaded "voice-changer-windows-nvidia-b2332" I opened the foler and all I see is" force_gpu_clocks | reset_gpu_clocks | and MMVCServerSIO"

Would I just run force_gpu_clocks?

[low shard]

open the mmvcserversio then the .exe

[valid spindle]

Is the voice changer supposed to be on a google tab?

Also sorry for asking a lot of dumb questions I know you probably deal with lots of people asking the same stuff😅

[low shard]

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

[azure patio]

is there any way to continue training if i have 150 epochs .pth file?

[next vortex]

So what app I should use without gpu needed and free?

[low shard]

you can't expect to run ai on potato hardware

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

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

which do u need

[next vortex]

RVC First

[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, broken right now)
• 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, broken right now)
• 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

[next vortex]

Could be a way I only use without do anything?

[low shard]

what?

you have to ofcourse do something to run programs

[naive harness]

What are epochs, batch size and steps?

[woeful bronze]

for one specific model I want to use on w-okada I get this message and I don't know whats causing it or how to fix it. can anyone help me?

Torch not compiled with CUDA enabled
torch\nn\utils\weight_norm.py:134: FutureWarning: `torch.nn.utils.weight_norm` is deprecated in favor of `torch.nn.utils.parametrizations.weight_norm`.
  WeightNorm.apply(module, name, dim)
Torch not compiled with CUDA enabled
torch\nn\utils\weight_norm.py:25: UserWarning: The operator 'aten::_weight_norm_interface' is not currently supported on the DML backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at C:\__w\1\s\pytorch-directml-plugin\torch_directml\csrc\dml\dml_cpu_fallback.cpp:17.)
  return _weight_norm(v, g, self.dim)

[low shard]

no

chunk controls the delay

higher chunk means more delay, and the chunk value depends on how much ur gpu can handle

[low shard]

elaborate:

• your pc gpu
• what u want to do
• what tutorial link did u follow

[woeful bronze]

• I have an amd (rx 7800 xt)
• I downloaded a model from weights.com and uploaded it normally to w-okada
• I followed a yt tutorial but other models I downloaded from this discord work fine - it's just this one model that gave me the error message

[low shard]

yt tutorial
that's part of the issue

youtube tutorials are extremely outdated

never follow youtube tutorials for ai programs

ai programs change alot and videos can't be changed like written guides

@woeful bronze uninstall everything you got off youtube

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[low shard]

read up the first link, the wokada deiteris fork is way better than the old original wokada you got off youtube

@woeful bronze also send the link of the model

[woeful bronze]

okay I will try, thanks!
https://www.weights.com/models/clueaogo700x7d0gcrylq938x this is the link

[naive harness]

Pls 🥹

[low shard]

the mmodel should be fine, lmk

[tough fiber]

anyone know this thing

@low shard sorry for ping really. im bothering u. if u available can u help me

[prisma mist]

"the paramters of the pretrain model such as the sample rate or architecture do not match the selected model" what do I do

[azure patio]

guys is it possible to continue training in another runtime?

[icy ember]

I am back after my PC reset, current setting are these

[olive basin]

does anybody how can i convert h5 format model into pt or pth

[low shard]

elaborate:

• your pc gpu
• what you are doing
• what tutorial link did u use

[low shard]

it's said in the message

it doesn't match the same parameters

[low shard]

f0: rmvpe

Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)

force fp32 mode on can cause better quality and stability at the cost of some delay

[prisma mist]

The parameters of the pretrain model such as the sample rate or architecture do not match the selected model.
Error: Pretrained model sample rate (48000 Hz) does not match dataset audio sample rate (40000 Hz).

It doesnt make sense cause my pretrained model sample rate is 40k not 48k

[low shard]

what pretrain are you using

[prisma mist]

TITAN by blaise-tk

[low shard]

and you're sure you're using the 40k sr version? show also your applio settings (since i'm guessing you're using that)

[prisma mist]

yes im 100% definitely using the 40k sr version, gimme a sec I'll show you my settings too

[marsh acorn]

hi it says the deiteris fork is a malware, also isn't it outdated since it came out in 2024 ?

[prisma mist]

After trying a couple stuff out, apparently all I needed to do was insert my dataset into audacity and export it as 32k SR, just for anyone facing that problem in the future

is it normal for the preprocessing to take that long?

[prisma mist]

Nvm, needed to restart Applio

[knotty moth]

before that, I'd recommend deciding the sample rate according to the spectrogram cutoff like following

https://cdn.discordapp.com/attachments/1159290139609137264/1344111265253032017/image.png?ex=6831c26a&is=683070ea&hm=e44466c1e69787e64cf1c5833f4f63b4c416ecc713312305337c3e7c53c05e90&

second, please double check the pretrain and sample rate chosen in the settings

at last, TITAN is rather old and has kind of quality issues, I'd recommend using the default pretrain or try: https://discord.com/channels/1159260121998827560/1339155300720054316

[prisma mist]

yeah thats how I decided to do 32k SR

[prisma mist]

oh damn, im already training, should I stop it and use the default pretrain then

[knotty moth]

test the model then see

[knotty moth]

more likely the program has closed or crashed for some reason (pls check the console)

instead of colab, try the kaggle one
https://www.kaggle.com/code/suneku/voice-changer-public

[fluid birch]

Hey all! I'm attempting to run Applio RVC but I keep getting an error reading: 'NoneType' object has no attribute 'pipeline'. Any idea what this means?

[knotty moth]

try downloading the precompiled release instead of getting headache on the manual installation

[sturdy sage]

so im runnin the appolio kaggle ngrok combo and i uploaded my audio file but i cant copy the path

whatever i do it just shows up as this

Preprocess completed in 0.00 seconds on 00:00:00 seconds of audio.

is my audio file wrong or somethin

or should it be a zip

im not sure i need help :(

[lost egret]

Timer: 00:03:17/content/voice-changer/server/HVoice.py:3: DeprecationWarning: The distutils package is deprecated and slated for removal in Python 3.12. Use setuptools or check PEP 632 for potential alternatives
  from distutils.util import strtobool
Traceback (most recent call last):
  File "/content/voice-changer/server/HVoice.py", line 10, in <module>
    from downloader.SampleDownloader import downloadInitialSamples
  File "/content/voice-changer/server/downloader/SampleDownloader.py", line 12, in <module>
    from voice_changer.RVC.RVCModelSlotGenerator import RVCModelSlotGenerator
  File "/content/voice-changer/server/voice_changer/RVC/RVCModelSlotGenerator.py", line 4, in <module>
    import torch
ModuleNotFoundError: No module named 'torch'
WARNING:pyngrok.process.ngrok:t=2025-05-24T06:37:01+0000 lvl=warn msg="Stopping forwarder" name=http-35747-ef82ddb1-a427-4741-86ed-c927acab8040 acceptErr="failed to accept connection: Listener closed"
--------- SERVER STOPPED! ---------
Timer: 00:03:18

using google colab "Hina Modified" version and idk im getting this error

[valid spindle]

Hi I downloaded voice meter and lighthouse and I did everything the guide said to do and now I cant hear audio, Do you know where I went wrong?

[knotty moth]

make sure to place a folder containing dataset wav file(s) in <the Applio folder>/assets/datasets

if it can't read the audio files, re-export them using Audacity or another editor without including metadata

[valid spindle]

Nm am I allowed to send screenshots in here?

[knotty moth]

!give-media-perms 30m @valid spindle

[valid spindle]

This is what all my settings look like, Does it look like anything is wrong?

[sturdy sage]

i lowkey dont know what the hell i was doing wrong so i just kept on using different thing but eventually i got it working on kaggle mainline

[knotty moth]

try VAC lite instead of vb-audio https://software.muzychenko.net/freeware/vac470lite.zip

you should also check and show the voice changer settings

[valid spindle]

Is the guide for this on the Deuteris fork guide?

[swift thunder]

Does anyone know what's going on with huggingface.co? I think it's down.

[restive kiln]

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[golden walrus]

anyone know how to improve model's tone ? because somehow i sound very monotone

do i need a set of data that the speaker have wider vocal range in order to achive better vocal range ?

[crude flame]

Either the inference audio is monotone or the dataset is monotone

[golden walrus]

[crude flame]

Yep

[analog obsidian]

train expressive data

[hallow thistle]

There is. Look up this link. https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#virtual-audio-cable

[crude flame]

Expressive data my goat 🐐

[analog obsidian]

[golden walrus]

let me google what is expressive data

[analog obsidian]

audio of someone using their whole vocal range instead of mommy asmr

[golden walrus]

also i want to ask about, is it okay to train a model then let it make tts data. and combine previous data and tts data together to improve my language pronounciation

i still can't spell right

i mean, i took data from a podcast. i think it is asmr

but okay, i think i get it

[crude flame]

Synthetic data is a no no for rvc

[golden walrus]

oh no, then do you know a way to get better pronounciation in different language? because cvec trained mostly on English data

[analog obsidian]

a new embedder is needed with a pretrain trained using that

[golden walrus]

dang it, can't find hubert base for Vietnamese anywhere

[limpid tinsel]

where i can find the collab of w_okada

or whatever is called

[hallow thistle]

What is your PC GPU?

[limpid tinsel]

Nvidia gt 710

thats why i am trying to get the google collab

i long ago used it in google without gpu, but now that i have one want to try again

[sly urchin]

whats the link to download w_okada?

[hallow thistle]

Yeah, no need to run W-Okada locally with this GPU. There are two W-Okada Colab notebooks, however, the original one is broken, while the fork one can get you terminated if you have free Colab.

[hallow thistle]

What is your PC GPU?

[sly urchin]

RTX 3070

[hallow thistle]

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

[limpid tinsel]

what does terminated mean, so i cannot use the voice changer?

[hallow thistle]

"Terminated" in Google Colab means you won't be able to use the Colab ever again with your Google account.

[limpid tinsel]

oh..

i see

[hallow thistle]

Use Kaggle instead, but you'll have to register with your phone number on this one. Here's the guide. https://docs.aihub.gg/rvc-voice-changer/cloud/w-okada-kaggle/

[barren gull]

Latency stats and metrics are typically based on GPU and CPU right now, right?

(aside from the code actually being good)

[naive harness]

What are epochs, batch size and steps?

[barren gull]

I'm not super familiar with all the terminology with this stuff atm — but I'm thinking in terms of real-time voice changing. Let's say I use something within the realm of 300 epochs.

I'm familiar with voicemod and w-okada, but curious if anyone is getting near-0 latency in realtime yet?

(obviously, you'd need a beefy GPU I'm guessing)

[misty vessel]

where can i download voice changer is there any tutorial? gpu nvidia 5080 cpu amd ryzen 7 9800x3d

[odd shale]

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[odd shale]

Check the Deiteris Fork guide.

[misty vessel]

is there one with video?

a tutorial with video

[odd shale]

Nope.

All video tutorials on YT are/and get quickly outdated so we don't recommend following these.

Just read the Deiteris guide.

[upbeat carbon]

quick question: can you record voice messages on rvc gui?

[celest dragon]

hey, Im new to this AI music world, where should i start if i want to make a rap song with the voice of a certain artist? most youtube tutorials are outdated or get deleted...

[prisma mist]

my voice model works fine in realtime but sucks badly when converting a voice to another, whats the reason for that

is it a problem with the dataset? do I need a more expressive one?

[simple ore]

people usualy have a quite opposite issue

[raw charm]

Hey everyone, I want to make a peter griffin tts video. what would be best fit for this? i wanna run it locally.

I tried F5TTS and technically it works but i have to reupload the sample every time. is there one i could use on of the models in #1175430844685484042 ?

[misty vessel]

can i ask u a question do i need to do both ir only nvidia rtx 5000 series i got a 5080

[odd shale]

Download only the RTX 5000 version in that case.

[misty vessel]

i dont know if i can ask this here but is there a tutorial for deep live cam?

[languid cliff]

So if i understand the guide correctly: the perf (number) is how good the voice performs, so you want to adjust the chunk as low as possible, while maintaining the perf number at green? As long as the perf is stable and green, does that mean the voice quality will be just as good at 100ms as at 5000ms?

[rich torrent]

I just need to know what website I can use to make a song then use the ai bot thing to add juice wrld too it.because the Suno ai music generator sucks

[misty vessel]

i dont know if i can ask this here but is there a tutorial for deep live cam?

[odd shale]

Ehh.. I fear not

[rich torrent]

Can I get help?

[rich torrent]

Help please

[olive plover]

what this why error??

FORK

[misty vessel]

is it normal that it opens up in a website?

the voicechanger

[odd shale]

Yup, that's normal

[odd shale]

Do you mean Deiteris fork, right?

Because that's a W-Okada fork

And don't worry, if it opens in a google window it doesn't mean you need internet.

[misty vessel]

yes with deiteris fork but can i still use it in dc?

[odd shale]

Yup, of course you can use it on discord

[abstract anvil]

Pretty sure that happens when you try to open the VC again when it's already open.

[gloomy mortar]

getting this error all of a sudden when trying to use the voice changer on kaggle

any fixes?

i tried to switch gpus on kaggle

but it js keeps saying "adding data"

[mystic narwhal]

how do you use the voice changer on overwatch

[sweet token]

Hello! I'm looking for the XMem.pth model (video object segmentation). All the official links are dead (Google Drive, HuggingFace, Mega). If anyone has a mirror or can re-upload, I’d be very grateful. Thanks!

[simple ore]

do you have an old huggingface model link?

[sweet token]

https://huggingface.co/akhaliq/XMem/resolve/main/XMem.pth

[sweet token]

Here is the original HuggingFace link (dead now):
https://huggingface.co/akhaliq/XMem/resolve/main/XMem.pth
If you have a backup, that would help a lot, thanks!

[simple ore]

it would not be something like https://huggingface.co/spaces/VIPLab/Track-Anything/blob/main/XMem-s012.pth ?

[sweet token]

Yes, exactly, it’s for XMem-based tracking.
Thanks, I’ll try this one out as a replacement for the dead original!
(If you have the old XMem.pth as well, I’d still appreciate it, but this looks good.)

[simple ore]

unfortunately the repo is gone, otherwise it would be super easy to get the file from the git history

[sweet token]

Thanks a lot for your help!
That new link saved me, much appreciated 🙏
And if you ever find the old XMem.pth, let me know — but this is already perfect!

[knotty moth]

the voice better be more expressive like in vtuber stream sessions to cover more pitch range

singing with enough pitch range is also fine

[knotty moth]

ye it's not bad for reading bedtime story ASMR, but not for expressive talking or singing

[dense rampart]

hey im trying to use w okada on my 3060 and the delay is insane, can someone help me?

[grizzled cove]

I'm trying to download SayanoAI/RVC-Studio but see no download button for the full repo, nor a github link. It's supposed to be a Local RVC. How do I download or use it? And can it run on my 5070 Ti and Ryzen 9 9950X3D?

[brittle wing]

Would forktime voice changer damage my high end laptop if used? Is it safe to use?

[knotty moth]

yes if you keep it overheat all the time instead of applying power limit and better cooling as needed

it should have less gpu usage than demanding games on an RTX 4090 laptop

[knotty moth]

it is kinda old and discontinued

try this instead https://docs.aihub.gg/rvc/local/applio/

[knotty moth]

what game are you trying to run?

also you can try using WASAPI devices for less latency https://rentry.co/LessDelayWasapi

[grizzled cove]

Thanks, will try it.

[knotty moth]

also note that rvc utilizes more on the GPU, so your semi-workstation grade cpu may be underutilized

[royal gull]

Is there any guides on how to link the voice changer to audacity??

[restive heath]

you want to use audacity to record the voice?

[royal gull]

Yes

[knotty moth]

for just recording, the non-realtime RVC for pre-recorded ones is always better

[royal gull]

I will definitely look into that!

[autumn cargo]

can someone tell me what does "torch" mean in my console when i run the w-okada voice changer

[simple ore]

pytorch is a machine learning library that is repsonsible for running rvc model

[clever onyx]

Hi, does the W-Okada fork (better version) fix the stuttering issue. I've been using the usual W-Okada with GeForce GTX 1080 ti and now currently with RTX 4070 and I'm still experiencing the stutters.

[knotty moth]

wdym stutters? in some game you were playing? if so that's an issue mostly plaguing unreal engine 5 games

[clever onyx]

By stutters I mean you can hear the voice lagging a lot. It lags even while I'm not playing games or in a call.

[knotty moth]

• check the settings, if too low try increasing chunk while keeping extra at most 2.7 s
• if the voice keeps cutting despite reasonable performance, try lowering the silence threshold
• you can try using WASAPI devices for less latency https://rentry.co/LessDelayWasapi

[clever onyx]

I lowered my silence threshold, it helped quite a bit, but still stutters/lags. My chunk is 512 (1365.ms, 65536) Extra is the highest so I can get the best results for voice, I will try lowering it.

[clever onyx]

Do you mind if I dm you and ask you for some more info about this?

[simple ore]

make a thread here https://discord.com/channels/1159260121998827560/1192011222023950368

[narrow portal]

I'm working on a voice model with voice clips ripped from a game

The thing is, almost all samples (which in total are over 5 hours but I'm just using 20 minutes in this case) are 22k, but a selection of 5 samples (with a total of 15 seconds) are 44.1k

So that means I should train my model on 32k or 40k sample rate?

Or should I just exclude the 44.1k samples from the dataset?

Thanks in advance

[simple ore]

5 hours is way too much and 15 sec is not enough, so pick best 15-30 min and use that at 32k

[narrow portal]

Should I keep the 44.k samples or should I exclude them from the dataset?

[knotty moth]

how is the perf indicator?

[clever onyx]

What is a perf indicator?

[oblique viper]

fuck this

please help

[knotty moth]

in the voice changer ui

[knotty moth]

and how about the cpu & gpu usage?

also try closing/disabling/debloating programs/background processes that may steal performance

[clever onyx]

Tried already, doesn't help.

I'll send a screenshot to you in private because I don't know what it is.

[knotty moth]

no, how about the cpu & gpu usage?

[oblique viper]

this code is stuck on running forever, without requesting me for the google drive access, can I get help?

[clever onyx]

It's my gpu graphics card

RTX 4070

12gb

[knotty moth]

bruh what % usage I mean?

[knotty moth]

!give-media-perms 30m @clever onyx

[clever onyx]

Oh thanks

Here

[clever onyx]

Oh

1 min

[knotty moth]

crepe is the problem, it's like tanking pathtracing performance in native resolution

go switch to rmvpe

[clever onyx]

20% cpu and 48% memory in my task manager

and I'll check gpu now

[oblique viper]

is using rmvpe on realtime voice changer fine if you've trained your model on crepe?

[clever onyx]

gpu is like from 5% to 30%

[knotty moth]

crepe in realtime is too slow as I said

[oblique viper]

yeah, I mean when you train a model, you can train it using crepe/rmvpe, but then I see there's also an option for which to use (rmvpe vs crepe) when on realtime voice changer

so I'm asking if it matters whether you use rmvpe on realtime voice changer using a model that was trained on crepe

[clever onyx]

It helped, thanks.

I appreciate it.

[simple ore]

popups disabled?

[oblique viper]

I fixed it, but I'd appreciate if you can answer my other question

[oblique viper]

@simple ore

[simple ore]

f0 is used for both training and for inference

[autumn cargo]

hey im using the w-okada voice changer but this version vcclient_win_std_2.0.78-beta laggs tf out of my pc

[simple ore]

it clearly says it gives your pc an std

[autumn cargo]

whats an std???

[simple ore]

try a lil more up to date software

-realtime

[autumn cargo]

alr

[simple ore]

https://rentry.co/forkvoicechangerguide

but it wont fix the problem if you got potato gpu

[autumn cargo]

ok

[full drum]

i dont know if the voice changers are different but whenever i start it using the start_http file it opens in command prompt instead of opening in python, anyone know why?

I have this one "MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a"

[autumn cargo]

thats how it runs i think

[full drum]

ohhh ok i just downloaded the one from this site, i thought it was supposed to open in python lol

[oblique viper]

can someone help me with cloud training?

I preprocessed my audio and everything shows that it's successful and I mounted to my google drive

but no new folders are showing up on my google drive

[warm garnet]

is linux optional, or do i really need it for w okada?

[oblique viper]

I mounted my google drive, but no new folders or files appear in my google drive when I try to preprocess and do all the steps

[restive heath]

windows 10+ mac linux all works i think

[oblique viper]

can anyone help please

[hallow thistle]

Linux is optional. W-Okada also works best on Windows 10 and up.

[warm garnet]

what should i change to prevent the delay?

[hallow thistle]

I'm not sure which chunk and extra number would work with AMD Radeon RX 580. But let say, the perf number is around 337 and is green at top left, so chunk may be around 390 ms, while extra is 2 s as it is. https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#known-working-settings-for-chunk-and-extra

[warm garnet]

okay

[hallow thistle]

While you can lower extra number so the perf number could be lower and less GPU usage, W-Okada would output lower audio quality at the same time.

[simple ore]

you're trying to use some ancient shite

[full drum]

oh yes the video i got it from is from 2023

[simple ore]

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[full drum]

so wait i just installed python for no reason or would i still have needed it

[simple ore]

the fork includes everything it needs

[full drum]

ohh aightt thank you bro 👊

[hallow thistle]

Don't believe any tutorial video on YouTube about W-Okada the realtime voice changer. They only telling you to install the original version, and are all old now. Use fork W-Okada instead for better performance.

[hallow thistle]

No need to do that if you wanna use W-Okada generally. All W-Okada versions have their own Python pre-installed.

If you have unknowingly installed a separated Python program, that's fine, although you may wanna uninstall it from your disk. Python program is useful if you wanna develop your own Python script or any AI-related program from start.

[dusty merlin]

i need help

with

fastsdcpu

not working

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[dusty merlin]

fastsdcpu error phyton not installed i installed 3 of phyon newest and 3.10 and 3.11

[hallow thistle]

A lot of people forgot to set PATH during the setup of Python, but there's a way to have it work after setup. The fastsdcpu mostly relies on your Windows' PATH variables in order to work, which includes a Python program you installed. On Windows 10 or 11, open Settings > About > scroll down and click on "Advanced system settings". On System Properties window, go to Advanced > Environment Variables > look at System variables section and click edit on "Path" Variable. This is where you add paths to the programs you wish CMD to run specific programs.

[dusty merlin]

which path tho

where to i put path

i still got them installed

[hallow thistle]

Now to find a path to your Python folder, it's something like C:\Program Files\Python310 or C:\Users\your username\AppData\Roaming\Python\Python310.

The your username is your current Windows user account you're running. Mine is named Lucy, so mine is C:\Users\Lucy\AppData\Roaming\Python\Python311.

[dusty merlin]

mys is nnn

[craggy saffron]

i feel like when u switch the output to your headphones its fast and not really delayed but when its on cable output and ur listening to it feels like cable output its making the okada delayed

[dusty merlin]

i hate using thecable out put i want it to make it voice live

[cosmic rampart]

how can i get w okada voicechanger

[dusty merlin]

i can help you

[cosmic rampart]

k

[dusty merlin]

dms

[cobalt carbon]

can u help me too

when i try the ngrok thing it gives me error

ModuleNotFoundError: No module named 'pyngrok'

like this

[dusty merlin]

uninstall it?

diasble anti virus

maybe install phyton

[cobalt carbon]

its not on my pc

im saying google colab thing

[dusty merlin]

oh

dms

[cobalt carbon]

someone pls help with google colab

ERROR: Could not find a version that satisfies the requirement faiss-gpu (from versions: none)
ERROR: No matching distribution found for faiss-gpu

[low shard]

elaborate

what's ur pc gpu? what do u want to do? what google colab link are u using?

[cobalt carbon]

my pc gpu 1650 so i want to use google colab

https://colab.research.google.com/github/w-okada/voice-changer/blob/master/Realtime_Voice_Changer_on_Colab.ipynb#scrollTo=lLWQuUd7WW9U

[low shard]

that gpu is good enough, you don't need to use cloud

are you planning to use a realtime voice changer while in game? if so, which game?

or just discord?

[cobalt carbon]

cs2

[low shard]

also, that's old original wokada

[low shard]

it's better you try locally first

cloud gets broken easily and has time limitation

[cobalt carbon]

i dont think its gonna work well cuz the delay

[low shard]

you should be able to run it fine with low graphics

[cobalt carbon]

no

[low shard]

Wokada has 2 main versions:

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

each version has it's own updates

the latest deiteris fork has way better performance and quality than the latest original

[cobalt carbon]

it just like runs 70-90 when using

[low shard]

wokada deiteris fork has way better performance

don't follow video tutorials

[cobalt carbon]

hmm

[low shard]

video tutorials use an old original wokada

[cobalt carbon]

can you recommend me a model

and send the fork link

[low shard]

uninstall whatever you got off youtube, get https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/

even suggested models are in the tutorial

[languid cliff]

70-90 perf? isnt that pretty good while gaming?

[cobalt carbon]

nope

for cs2 its bad

[languid cliff]

ah

[vestal gust]

hi guys the audio like cuts in an out all the time and it just always sounds bad? anything u rec to help me

[cobalt carbon]

can u tell me a good model

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#models-to-try

[low shard]

elaborate:

• your pc gpu
• what u want to do
• what tutorial did u follow

[low shard]

@languid cliff that's actually good delay while gaming btw, don't expect 0ms delay, but i feel like you were using an old original wokada version that didn't even show perf and you're just confusing values

[cobalt carbon]

can you tell me which settings are good for low end gpus

[vestal gust]

i have a gtx 1650 4GB, i wanna do real time voice changer, i followed a guy on youtube who does like e girl trolling on discord name was duckus

[low shard]

https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#known-working-settings-for-chunk-and-extra

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

[brittle wing]

hello can i have some help

[low shard]

that's already your issue, all video tutorials are outdated

that tutorial is like more than a year old

in ai field, it's prehistorical

[vestal gust]

true

[languid cliff]

yeah, even with a 5090 i get about perf 100 in some games, but thats if the GPU is on full blast (80-100% usage)

[low shard]

@vestal gust uninstall everything you got off youtube, are u going to use it in games?

[low shard]

yeah, he prob just confused values lol

[vestal gust]

yeah, only really fortnite, roblox discord

[brittle wing]

hey uhh this is the support chat right can i get some help?

[low shard]

elaborate

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[low shard]

fortnite not so sure, but the rest should work fine

uninstall everything off youtube

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[low shard]

read the 1st link

wokada deiteris fork

[vestal gust]

ok cool

[brittle wing]

im using w okada and am new to this whole thing but the voice is very very choppy and spouts gibberish. when i do the passthru thing i can hear my normal voice properly tho

[low shard]

what's your pc gpu? what do you want to do? what tutorial link did u follow?

[brittle wing]

i think mx330 idk if its enough i dont know alot about tech

this is the tutorisal i followed https://www.youtube.com/watch?v=KkDjqteIqBE

[vestal gust]

for a 1650 would u reccomend online one or not? @low shard

[brittle wing]

i jusr wanna be able to use the voice chnager and have it not be choppy

?

[low shard]

that's weak asf

[low shard]

also, that tutorial uses an old original wokada version

the only way for you is to either buy a better pc or use cloud (remote good pc) with limited time

[brittle wing]

so hwat do i do to be able to have a non choppy voice is that possible or

[low shard]

should work fine except for intensive games

roblox and discord are completely fine

id suggest locally

cloud has limited time and breaks easily

[brittle wing]

so its because my pc is too weak? but im using the version without the gpu thing like the std thing thats what i downlaoded

[low shard]

first of all your pc is weak, second of all you're using an old original wokada which has even shittier performance

your gpu is just going to have shitty performance even with the latest wokada deiteris fork that has optimizations

[cobalt carbon]

@low shard bro how can i run this on my pc locally , it opens a browser tab

[low shard]

so either you buy a better pc or use cloud with limited free time

[low shard]

it opens locally, it just uses a local Web User Interface https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window

[cobalt carbon]

ahh okey

[brittle wing]

so iots because oif my gpu? but i heard some people without gpus are able to use it propeely how can thye do that?

[low shard]

you could technically do it with a cpu, should you? no, it will run like complete dogshit

it's because your gpu is weak + you're using outdated software

I told you your only 2 options

which one are you interested into?

[brittle wing]

ok then

i was thinking of buying a new laptop anyways

but then ewhat version should i install

[low shard]

if u wan to do it locally, wokada deiteris fork https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork, which version depends on the future laptop gpu you're going to get

[cobalt carbon]

shouild i select server

[low shard]

hopefully a laptop with an nvidia rtx decent gpu

[low shard]

give it microphone perms

what browser are u using

[brittle wing]

would a 3050 be fine?

[cobalt carbon]

brave

[low shard]

are you going to use it on games too? if so, which?

[brittle wing]

nah just for discord

[low shard]

try chrome/firefox/microsoft edge and be sure to give microphone perms

[low shard]

you can choose

client: easier and can use noise suppression, more delay

server: less delay, harder, can't use noise suppression

[low shard]

then it's fine

[brittle wing]

so 3050 is fine

?

[low shard]

yes, just don't expect 0ms delay since that's not even a thing lol

[brittle wing]

okay

wow this ai stuff needs alot of power lol

[low shard]

that's the whole point of AI

[brittle wing]

ig

[low shard]

you know right chatgpt uses a shit ton of way more expensive gpus?

[brittle wing]

oh i didmt know that

[low shard]

ai doesn't run off air

they all use a shit ton of gpus which can cost over thousands of dollars

every ai site you see uses shit expensive hardware, which is why they make you pay for it, because it's running on their pcs

[brittle wing]

ohhhh

[low shard]

chatgpt doesn't run on your laptop at all

you can also run wokada deiteris fork on cloud (remote good pc), but you will have limited free time, which is lucky considering how expensive gpus are

[brittle wing]

what is that

[low shard]

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

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

[brittle wing]

ok

ok then

thanks for the help

[low shard]

yw

[vestal gust]

just a quick one

is it supposed to open on the web?

like the local version

im assuming the more chunks the more quality right but more delay?

[cobalt carbon]

can you help me to make settings good , sometimes my sound is robotish

[gusty forum]

"Hello everyone, does anyone know which AI preserves the quality of a person's image? I mean, when generating an image with different effects, the person's face should remain as it was originally. However, both GPT and Grok distort it during generation. Which AI doesn't do this? Even if it's paid, it doesn't matter, please tell me about all the ones you know via private messages. Well, you can write here too, but it would be preferable if you could message me privately, thank you

[low shard]

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

[low shard]

no

chunks controls delay, more chunk more delay

extra controls a bit quality

[low shard]

send a screenshot

[languid cliff]

https://i.imgur.com/TqNkoZW.png anoyone else encountered this issue?

[vestal gust]

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

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

• **RVC GUI Mainline Realtime**

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

[sudden quail]

can anyone help me using copy of gui for rvc?

[viral mason]

anyone got a good reverb removal tool for really heavy reverb

[fresh cairn]

does deiteris fork no longer use a little UI?

its all browser?

[low shard]

use rmvpe without onnx, extr to 2.7, and be sure to get a good model

[low shard]

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

[low shard]

elaborate your issue

[low shard]

original rvc / mainline, is not so much updated anymore, and doesn't support the rtx 50 serie, you need to use a more up to date fork like applio

https://docs.aihub.gg/rvc/local/applio/

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

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

[languid cliff]

I tried with Applio too, and i got the same message 😦 I completely uninstalled torch and python, and tried installing the nightly Torch build, as i saw people recommend that. But didnt have any luck

https://i.imgur.com/XYQZptZ.png pretty much the same message

"import torch

torch.version
'2.8.0.dev20250524+cu128'

[low shard]

you sure you got the 3.2.9 version off https://docs.aihub.gg/rvc/local/applio/ ?

not sure if they officially added rtx 50 support, if they didnt:

How to (unofficially) use Applio for RTX 50 serie cards

Follow to download it as said it in https://docs.aihub.gg/rvc/local/applio/

After you extracted the precompiled, go to the path in Windows explorer, write "CMD" and press enter, then in CMD write env\python -m pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --upgrade --index-url https://download.pytorch.org/whl/cu128

If you get any already satisfied requirement issue, run env\python -m pip uninstall torch torchvision torchaudio then the command said above

[languid cliff]

3.2.9 yeah

[low shard]

@simple ore i thought rtx 50 support was already supported in applio 3.2.9?

[low shard]

try the method i told ya

[languid cliff]

That seem to have worked, no error message now!

[low shard]

you're welcome

[languid cliff]

Thanks

[simple ore]

no, 3.2.9 still uses torch 2.3.1

[low shard]

why

[simple ore]

I think we rushed a day with making the release, the pytorch release was a few hours after we made a build

there was a bug preventing training from working properly as well

had to make sure 2.7.0 fixed that

[languid cliff]

Last question, is there any guides anywhere on how to do optimal training?
Recommended batch size and epoch compared to voice length? And like what voice data do you NOT want, and what voice data do you want a lot of, etc?

(aside from the obvious one, background noise)

[sudden quail]

need a tutorial

[brittle wing]

Hi, how do I contact someone to remove some model posts that were mine?

[oblique viper]

all I can find is this

where did all my training go

please someone help

no way all of these hours I wasted are lost

dude this is so upsetting

I spent 2 days tinkering with this shit

what the fuck why is there no warning

anyone online

[simple ore]

those are not the logs you're looking for

if you had backup enabled, all your files should be in your google drive

[oblique viper]

that's the thing I did not have backup enabled

why the fuck are there no warnings anywhere for fucks sake

you know atleast somewhere a little warning that you're gonna lose all your shit just for being afk

all I saw was a subtle hint that I can do backups

[simple ore]

in Applio NoUI auto-backup is enabled by default

[oblique viper]

backups are usually for unexpected crashes, not a built in ass feature that gets rid of all your stuff

just wow

[simple ore]

you're using a borrowed cloud server to do your shit for free, why do you expect it to be permanent?

[oblique viper]

well atleast let me keep my last files before kicking me out

they could literally just have it disable the GPU and let me download my last files

[simple ore]

you're directing your anger towards a wrong target

[oblique viper]

there's nothing to direct my anger towards honestly

[simple ore]

yourself, for not learning what colab is, before starting using it

[oblique viper]

sorry for not checking that a cake has nails in it before taking it in a birthday party

silly me

should've expected nails

honestly everything made by google is smelly farts should've expected colab to be ass too

please atleast put a suggestion to have auto backup on in the aihub documentations so that less people have to go through these frustrations, because I'm 100% sure I'm not the only one who thought Google employees wouldn't be so braindead to have you atleast keep your files if you're afk for too long

or have it be enabled by default in the colab just like it is in Applio NoUI, because there's literally 0 reason for it to be off besides bringing you unneeded pain if you're new

[oblique viper]

okay so I have auto backups on, before I go to sleep, is there any other trap I should know about? No jester that will come to my home at 3am and delete my files?

[oblique viper]

and can I please get an explanation of why this keeps popping up?

same process, same everything

sometimes works

sometimes doesn't

I infact did not forget to slice the audios.

there errors are popping out of nowhere and for no reason dude

I can't stand this anymore

the guide is dogshit

[narrow portal]

Is it normal for the TensorBoard charts to be very "zig-zag"ish?

I can't send a screenshot cuz I don't have image perms

[oblique viper]

it's normal for everything to have a problem

if everything is going smoothly then you're doing something wrong

[oblique viper]

also turn on your auto backups trust me you'll regret not having it on

big secret

[narrow portal]

I'm doing local training and I'm saving every 25 epochs, so yeah

[oblique viper]

ok

I wish I could do local training

this colab is so ass

[narrow portal]

Yeah, Colab has a lot of limitations

That's why I moved on to Applio

[oblique viper]

unfortunately I'm stuck with an unsupported amd gpu

[narrow portal]

Okay, now I can send images, so

[simple ore]

what gpu?

[narrow portal]

This is what it looks like so far

[oblique viper]

6700xt

[oblique viper]

this looks fine, you can up your smooth settings

[narrow portal]

Oh, right

I forgot to change that

Okay, now that looks way better

[oblique viper]

it took me like 30 minutes to setup local applio

[simple ore]

why do you think it is not supported?

https://docs.applio.org/applio/getting-started/installation#amd-gpu-support-windows

[oblique viper]

this

[simple ore]

dont smooth avg50 charts, they are smooth enough already.. 0.5 max

[simple ore]

follow the guide I gave you

[oblique viper]

do I not need support for HIP SDK?

I have no idea what that is

[narrow portal]

Oh, alright

[narrow portal]

Strange, since the AI Docs guide tells me to use 0.987 smoothing

[simple ore]

that's for old logs that were saving values at the end of the epoch and were quite random

[oblique viper]

https://tenor.com/view/mercy-overwatch-come-here-gif-14280244

[narrow portal]

I reverted back to 0.5 and it looks zig-zaggy again

[oblique viper]

those zig zags are fine

[narrow portal]

Oh, alright then

[oblique viper]

you can better observe the lowest points

[oblique viper]

thank you, I don't wanna touch that cursed colab ever again

[simple ore]

6700xt is faster than colab

[oblique viper]

yeah I know

I followed the tutorial step by step

[simple ore]

faailed to set path

[oblique viper]

what else do I need to set path to

installed wrong hip sdk version 1 sec

[simple ore]

delete zluda folder, use a correct patch

[oblique viper]

the one here?

I've already did the other steps with the wrong hip sdk version, do I have to do anything about that too

[simple ore]

5.7 zluda wont work with 6.1 sdk and other way arouund

[oblique viper]

god dammit my brain is so fried after countless hours of bashing my head against this

I'm deleting the applio folder and reinstalling it and redoing all of these steps again

hahaha... again...

it's 4am, I have a doctors appointment at 10am

this is awesome

@simple ore do I skip this step for the applio installation if I plan to install zluda?

I guess I do

is this something I need to worry about? @simple ore

I installed all over again, did every single step as by the tutorial, yet I'm getting this same problem again

what am I doing wrong? Please honestly tell me what I'm doing wrong

it's hard not to blame the guide if literally following it step by step causes issues

[simple ore]

you seem to be mixing two guides instead of just using the one I gave you

why

[oblique viper]

that's not two guides being mixed

that's installation of applio itself

[simple ore]

you take compiled applio version, unzip and that's it for applio

[oblique viper]

that's what I did..

[simple ore]

then you follow the guide step by step

[oblique viper]

that's exactly what I did

[simple ore]

verify the path

[simple ore]

okay

did you run patch zluda 57 bat?

[oblique viper]

yes

[simple ore]

from Applio's root

[oblique viper]

what does from Applio's root mean? I opened the .bat file inside the applio folder

[simple ore]

okay

now you run run-applio-amd.bat

[oblique viper]

that's that

I feel like I'm going insane, what did I do wrong

[simple ore]

hm... what driver you're using for AMD card?

25.5.1 is fucked up

[oblique viper]

where can I check?

[simple ore]

open the adrenalin control panel thing

try this instead, it downloads the latest version compatible with 25.5.1

[oblique viper]

I think I had none, I am currently installing adrenalin

it asked me if I wanted to download 25.5.1 or 24.Q4, I chose the downgraded one and installed it

[simple ore]

what asked?

hip sdk may prompt to install and old version

[oblique viper]

the adrenalin software thing, it wasn't installed on my pc

I had no drivers

for the hip sdk do I install the 23.Q4 version it asks me to?

[simple ore]

no, 25.5.1 is fine with that bat file above

[golden shadow]

yo whats up guys

hows yall day going

[simple ore]

23.Q1 is a pro driver without any game support

[oblique viper]

I don't really know what a pro driver is

[oblique viper]

oh

[oblique viper]

@simple ore is that normal?

I reinstalled everything once again, this time while having 25.5.1 as a driver, I did everything exactly step by step, the only difference being is that I used your patch-zluda-hip57.bat file

this is exactly what I was fearing I'd have to go through if I didn't use google colab..

[simple ore]

torch installed as well?

at this point you may as well check that you did not skip vc++ runtime

[oblique viper]

I fully uninstalled all of torch stuff now, uninstalling vc++ runtime and restarted pc, I'll try to make this installation as clean as I can

if that doesn't work then at that point I might aswell clean install all of windows lmao

[simple ore]

you just need to follow instructions step by step without skipping anything

because we may very well find out you did not replace libraries next

[oblique viper]

it

is working

it's almost 6am

I would gift you nitro if I had the money

[narrow portal]

Got this thing while training, idk what to do now

[oblique viper]

I'm too cool to censor my user name

[simple ore]

what are you using?

[narrow portal]

Applio

I left it training in the background while watching YT

[simple ore]

gpu i mean

[oblique viper]

@simple ore

I give up

same errors I was getting on google colab

[simple ore]

not important

the big error dump is more important

[knotty moth]

there was some issue in the feature step (the init_thread spamming stuff)

[oblique viper]

this is where it begins

[simple ore]

use amd cleanup utility, or/and DDU, remove whatever driver you have installed, install adrenalin 25.3.1

it should not fail on the kernel compilation

the thing about the index is unrelated.. it simply cant find any extracted files in your model folder

[oblique viper]

I'll wait another ~10 min until it finishes with the "compilation in progress..." thing again

[narrow portal]

Oh

NVIDIA GeForce GTX 1650 (4 GB)

[simple ore]

you dont have enough vram to train anything

[narrow portal]

I actually trained a few models in the past and they worked

But only when I leave my PC training without doing anything else

[simple ore]

it requires 4.5GB as base + more per batch number

you may very well train on CPU

[narrow portal]

As I said, training only fails when I'm watching YT, playing games or other stuff while training locally

It trains perfectly fine when I leave my PC without doing anything else on it

[narrow portal]

Is an Intel Core i5 enough?

[simple ore]

core i5 from 10 years ago? probably not

[simple ore]

you're using shared memory which drops the already low training speed to 1/3-1/5 of it and of course it fails when you try to use vram with a browser or discord using hardware acceleration

[narrow portal]

Makes sense

I'm completely broke rn so I can't afford a new PC to replace this "pseudo" gamer TUF Gaming notebook I have

So I can only work with what I have and nothing else

[simple ore]

you're a prime candidate for kaggle/colab 🙂

[narrow portal]

Aren't there limitations tho?

Like, a GPU time limit, or idk

It's been a long while since the last time I used Colab for training

Ever since the removal of free GPUs for Colab stuff, I stopped using it and moved on to local training using RVC/Applio

[narrow portal]

I'm gonna continue with the training but this time I won't use other apps while training

Also, was this error able to corrupt the training on my model or should I be fine?

[worldly bison]

guys why i cant use the voice changer app is there any latest app?

[knotty moth]

perhaps doable for batch size 2 (and with checkpointing option in Applio)

but not so recommended as the model would tend to be unstable

[narrow portal]

Right now I'm doing okay with batch size 4 and checkpointing

This is how the voice sounds so far with 87 epochs

[knotty moth]

up to 4 hours of daily colab GPU session or you can consider kaggle

[knotty moth]

ye it won't cause out of memory error but the performance is mostly bottlenecked

not a thing to worry since it's not a realtime workload

fyi batch 4 is recommended for less than 30 min dataset, okay-ish for 30-60 min depend on the dataset diversity

[oblique viper]

@simple ore

at this point it'd be easier to get hired, make enough money for and buy a good nvidia card

is there literally anything I can do without 90+ errors

[jovial patrol]

hello , i have a question , how to make rvc path with my own voice and how to make that auto detect index path , idono anything about it can someone help me with that ? please

[oblique viper]

Not enough data present in the training set. Perhaps you forgot to slice the audio files in preprocess?
atleast let me run google colab ffs

why even google colab can't work properly like what the fuck is this fix your shit for gods sake

[knotty moth]

make sure you have run preprocess and feature extraction without issues

[oblique viper]

I did

same process as I do locally, and I did the exact same process on google colab a day ago, with the same dataset, and it worked normally

but suddenly it gives me this error all the time

there's a higher power that literally doesn't want me training ai

[knotty moth]

I mean did it throw some errors or the results like: "preprocess completed in 0 seconds" and/or "no-feature-todo"?

I also suspect you're prob running an outdated/old colab like rvc disconnected

[oblique viper]

take a look for yourself

[oblique viper]

https://docs.aihub.gg/rvc/local/applio/
I'm running the link that's here (https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio.ipynb)

there's no way my problem is so out of the ordinary not even the helpers here know what to do

it's not like google colab would yield different results based off your personality or something

this colab has a personal vendetta against me I swear

[knotty moth]

how is the dataset structure and what preprocess option were you using?

[oblique viper]

I used the dataset creator option where I uploaded a 12 minute mp3 file that is just purely talking with no background noise, it then created a dataset folder with the 12m mp3 file in it and I used that as the path (it automatically puts in the path for you)

I used all the default options

for the pitch extraction algorithm I use crepe

[knotty moth]

well so it's in mp3 format, I mean how did you put the dataset?
make sure it is a folder containing an audio file (if mp3 may cause issues, try wav instead)

also make sure the preprocess and feature step don't take instantly

[signal spire]

Hi,

Does anyone here use Moises app? Have you guys ever tried to have an apk version of it? Hhuhuhu I'm enjoying moises now but I can't pay for it's premium price

[oblique viper]

if that helps you, I recorded a 3 minute video of the process I do from start leading to the error

[oblique viper]

(just to clarify, the error at 0:39 is due to me trying to put in the folder how you told me to, I usually put the mp3 file right away and it doesn't say error)

[hallow thistle]

You must be unlucky enough trying to get Applio on Colab to work and it still fails every time you do so. Have you ever tried the Kaggle one before? Or Kaggle still broken for you?

[oblique viper]

I have not tried Kaggle, I'll check it out

[hallow thistle]

As what you vented hard about it in #🧬│ai-chat , I'd say give up temporarily on this one. Since you have forced yourself too much trying to get Applio Colab to work. You can try again another time when you feel right.

[oblique viper]

this is not normal, I've tried google colab, it showed me the error I'm currently trying to fix, then after trying the same process again and again, for some reason it worked once and that's when I didn't have auto backups on

I tried installing zluda

and it's a whole mess in it of itself

[simple ore]

that's your fucking problem

[knotty moth]

• you haven't checked out Advanced settings in preprocess
• drag n drop a zip file, not a folder

[knotty moth]

I think he messed up on the preprocess

[hallow thistle]

Also, you have just created two threads in #1192011222023950368 hours ago, whereas these two refer to the same thing, but neither has an answer from any helper. Make sure to focus on either one you think it's more obvious for solution.

[oblique viper]

what's the problem with crepe?

[oblique viper]

zip file containing the dataset folder?

[simple ore]

it failed to create f0 files

why did it fail I dont know, try running inference with the same crepe parameters to find the errror

[knotty moth]

please check the folder Applio\assets\datasets\<model name> in the left pane

[simple ore]

unortunately the error on training screen is not visible

Applio does not use .zip for preprocess

it is a path to the folder with a file(s) (.wav, mp3, etc)

not a path to the file

[oblique viper]

so what I did was correct there?

[simple ore]

yes, just use rmvpe

i'll check colab for crepe thing

[hallow thistle]

Crepe is an old pitch extraction model for RVC from 2023, older than rmvpe. Although you can try rmvpe on this one, and see if it works.

Unfortunately, you deleted an earlier thread I replied in #1192011222023950368 for whatever reason, so I couldn't find my own answer to remind you again.

[little owl]

hi im trying to resume training thru rvc disconnected collab, but not sure if im doing something wrong. i used the same settings.

[oblique viper]

I closed the thread there because at the time the problem was fixed

[oblique viper]

its an outdated colab

[little owl]

ah okay is there another colab available??

[hallow thistle]

[hallow thistle]

Have you tried Applio?

[knotty moth]

applio has crepe with hop length

[little owl]

isnt that local only?

[hallow thistle]

No. Applio is also available on Colab and Kaggle, aside from being a local program

[little owl]

ah okay ill try it out. would i have to restart training complately or can i resume from my checkpoints i have on drive already?

[oblique viper]

you should be able to resume

[little owl]

ok perfect thank u

[oblique viper]

switching to rmvpe fixed it

[hallow thistle]

You can resume training on Applio if you have trained a model there before. But I'm not sure how one would tryna transfer training data from Disconnected to Applio.

[oblique viper]

the mongolian throat singing that I put in my background contributed to the fix too

[hallow thistle]

[simple ore]

what hop size did you use for crepe?

[oblique viper]

I tried 70 and 1

[simple ore]

so seems like PEBKAC problem

works fine when you dont do something stupid like that

[oblique viper]

please educate me on what those hop sizes are

because there all it said is that lower hop sizes do something better with pitch stuff

so I put lower hop size

[simple ore]

where does it say that?

[oblique viper]

here in the video I used default 128 hop size btw

[simple ore]

default 128 is fine

but not 70 nor 1

[oblique viper]

that didn't work either though, as you can see in the video

[simple ore]

see my screenshot - worked fine

[oblique viper]

see my video - didn't work fine

[oblique viper]

the small grey text above hop length says
"Denotes the duration it takes for the system to transition to a significant pitch change. Smaller hop lengths require more time for inference but tend to yield higher pitch accuracy."

if it was a hop length issue then it would've worked in the video where I did not change it

[knotty moth]

stick on either 128 or 64 for most cases

[simple ore]

neither works properly anyway and makes no difference

f0's target is 160 hop to match feature extraction

when you use 128 or 64 hop it creates a larger array that has to be interpolated down to len/160 frames

[oblique viper]

so I should set it to 160 if I use crepe?

[simple ore]

neither crepe nor rmvpe is ideal

[oblique viper]

what is ideal

[simple ore]

nothing yet

in 80% of cases all 3 extractors agree on the curve

the issue is that they disagree on 20% and it is a different thing

crepe hates noise and makes non-0 values

[oblique viper]

I see

[oblique viper]

is there a big difference between using a 10 minute recording vs a 30 minute recording for training? I feel like my index is a bit off on the accent and considering to use a bigger recording next time

[simple ore]

using index is what gives you an accent, because the voice model does not have the sounds that match the audio you're inferring

like german model would say 'ze' instead of 'the'

without using an index the accent goes not minimum

[oblique viper]

yeah I know, I'm curious if a longer recording would give a higher quality .index

if there's any significant difference in index quality between 10m vs 30m clip

[simple ore]

larger dataset results in a larger index, but unless it contains the sounds close to the target audio there's still going to be an accent

[oblique viper]

if you listened to both the smaller index and the larger index side by side, would you be able to tell a difference easily?

[simple ore]

btw, you can use any index with any model

[oblique viper]

the index that I need is quite specific

I doubt there's index's for an Israeli accent that I can find

french could work maybe

[simple ore]

i'm sure you can find some celebrity model and take its index

[oblique viper]

yeah I can find something, cutting out a whole bunch of talking clips sounds a bit like a pain but atleast it's not what I had to go through with zluda

[meager dome]

hi, is it possible to make the ai model read text? I cloned my own voice and I have a german project to do and I can’t speak german so this is why it s kinda important

[oblique viper]

there's a tts tab in the applio thing

[harsh oak]

Which models work best for audio restoration?
I've recorded alphorn in the mountains with my camera but I want to remove the annoying clicking noise from the focus motor

[oblique viper]

you might be able to use some free AI tool that's made for isolating lyrics and instruments

[wind cradle]

can anyone help me finding the ai software?

[plush raven]

Is there a way to make the tts in applio change it's tone by giving some kind of tag like {{whispering}} ????

[wind cradle]

can someone send me the download link for the voice changer please?

[oblique viper]

#1159513888199540817

[wind cradle]

thank u

[meager dome]

oh i had no idea

thanks

[harsh oak]

I used Denoise-Debleed by Gabox. That worked great!

[wide saddle]

hi whats a good ai photo enhancer

thats free

[gloomy grove]

есть кто то русский?

[simple ore]

no, use a different tts

[blazing mural]

Hello

[hallow thistle]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

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

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

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

__**BE PRODUCTIVE.**__ 🤝

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

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

[hallow thistle]

While not many people would own a Macbook to test and run W-Okada, it's rare from what I have seen, most people here got Windows and sometimes Linux.

Which Macbook do you have? There are two variants of fork W-Okada for Mac of different CPUs.

[blazing mural]

!howtotask

[hallow thistle]

No, that's not what I meant.

[blazing mural]

oh

I don’t know what one I have

[hallow thistle]

There's a way to check if your Macbook is M or Intel, which is somehow located in About This Mac menu. https://static1.howtogeekimages.com/wordpress/wp-content/uploads/2023/09/click-apple-about-this-mac.png https://macreports.com/wp-content/uploads/2023/04/compare.png

The "Processor" or "Chip" part would tell you the CPU name of your Macbook.

[blazing mural]

M1

Chop

Chip

[hallow thistle]

Apple M1 is Apple Silicon. Now have you download this W-Okada? https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/#download-mac-silicon

[blazing mural]

hi sorry me again!, when i downloaded the thing it said that one of the files were damaged

Not the slicnonn

The acual files

[oblique viper]

I have once again come back for help

I think?

I'm trying to continue my training off the backup, but it looks like something went wrong

my steps:

1. Mount drive -- success
2. Load backup with the accurate model name -- success
3. Auto backup on -- success
4. Install Applio -- success
5. Start Applio with gradio (like usual) -- error above

[simple ore]

seema you forgot to cone the app

then run install

[oblique viper]

what do you mean cone the app?

[simple ore]

clone

[oblique viper]

clone what? to where? I ran the "Load a Backup" option and it looked like it automatically created an Applio folder with the stuff in it

[oblique viper]

there is no such option

[simple ore]

then it is not possible

the UI colab should clone everything properly

[oblique viper]

never in my life did I see a "clone" option in google colab 😐

[oblique viper]

because whatever that is, it's not the colab I'm using

do I need to open a different colab for backups?

[analog obsidian]

https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio_NoUI.ipynb

[oblique viper]

ah, a different colab, as I suspected

[oblique viper]

thanks for the big red arrows that totally helped me see that in https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio.ipynb

like dude don't act like I'm supposed to know all this out of the box and be like "ah yes! For backups there needs to be a different no UI colab! So intuitive!!"

it's so obvious to know things that are not mentioned at all in the guide of course

ffs

and why does this exist ????

for no reason at all?

just because?

just so that I get bamboozled?

genuinely educate me

come on

instead of acting like I'm stupid for not knowing shit as a beginner that is not mentioned in the guide even once

[analog obsidian]

tbh idk why he was so mean towards you but yeah the docs aren't up to date and that specific colab is in applio's github site

[oblique viper]

okay, thank you

[analog obsidian]

im not affiliated with applio btw