✨│ai-help

[regal hedge]

CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.
Traceback (most recent call last):
File "voice_changer\VoiceChangerV2.py", line 239, in on_request
File "voice_changer\RVC\RVCr2.py", line 228, in inference
File "voice_changer\RVC\pipeline\Pipeline.py", line 161, in exec
RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

[simple ore]

well, there's train.log

means it is not saving model until the very end

[ember panther]

i mean there are but what should i do with it there isnt a model showing on the website

[simple ore]

batch size 40 -- you must be crazy

[ember panther]

what should i do .d idk that things

[simple ore]

unless you're training on 1000 hours of audio

well, check the logs folder for train.log

[ember panther]

i found it

[rare goblet]

Ayo? @ember panther level 2 !!!

[ember panther]

so what i gotta do with it

[simple ore]

read it?

[fathom kraken]

So I've put around 30 audio of 3 second of character quotes so they're kind of the same length, some being 6 or 4. The voice is okay (I know i'm not in the recommended 10 minute at all) I suppose the reason it cannot pronounce certain words is because they never pronounced "Fish" so when it has to mimick "Fish" it lags out?

[ember panther]

i mean that saying all keys matched successfully in the end

[simple ore]

and nothing else?

[ember panther]

i mean there are some more things like epochs but there isnt not some kind of fail message

here the image of it

[austere stream]

this has probably been asked many times before, but is this model overtrained? if so, where?

[raven ginkgo]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[rare goblet]

Ayo? @raven ginkgo level 2 !!!

[simple ore]

check fm, mel, kl charts

[ember panther]

is there any problem with the image of mine?

[simple ore]

for some reason the training did not start... try batch 4 instead of 40

[ember panther]

what does batch do

[simple ore]

number of samples to train in parallel

[ember panther]

alr i will try

[austere stream]

ive never checked these before while training, just g/total, so is it looking okay?

[royal juniper]

PermissionError: [Errno 13] Permission denied: 'C:\AI\Voice_Trainer\Mangio-RVC-v23.7.0\ONNX'

how do i solve this?

[fathom kraken]

Also was wondering, why can Elevenlabs make perfect voice replikas with 60s and not mainline, is it because they have insane processing power? Epoch? Models not publically available?

[royal juniper]

how do you solve permission error like wtf?

[simple ore]

fm looks scuffed, small data set?

[austere stream]

about 19 minutes so yeah

[simple ore]

check that save around 17k steps

[austere stream]

alright thank you

[royal juniper]

PermissionError: [Errno 13] Permission denied: 'C:\AI\Voice_Trainer\Mangio-RVC-v23.7.0\ONNX'

can somebody please help me with this?

likw why does everything is causing problems...

[simple ore]

installed / running as admin?

[royal juniper]

I can't run it as admin. if says that system can't find the path.

pls help I genuinely don't understand why everything just keep breaking. and why this thing just not working the way it's supposed to

it's one thing on top of another

[simple ore]

what's your GPU?

[fathom kraken]

I just tought of using elevenlabs voice cloning to create more samples for my character voice and then use it to train on mainline

[royal juniper]

rtx 4060 ti 16gb

it should be enough

[simple ore]

okay, so get a packaged RVC Mainline or Applio, forget this Mangio nonsense

[royal juniper]

what's the difference?

[simple ore]

https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/RVC1006Nvidia.7z or https://huggingface.co/IAHispano/Applio/blob/main/Compiled/Windows/ApplioV3.2.6.zip

mangio has not been updated for 15+ month

RVC is less stale, but there there are some bugs unfixed for ages

Applio is most up to date-ish

[royal juniper]

what is the difference between those 2

like applio and mainline

[simple ore]

overall they do the same thing and produce the same model / fully compatible

the difference is one project is receiving updates and the other is mosty on life support

and mangio is a dead corpse people keep digging up and trying to rescuscitate

[royal juniper]

so applio receives updates?

mainline is also dead?

[simple ore]

mainline authors have a new shiny thing to play for their master's degree or something

[royal juniper]

what's that shiny new thing? I won't have the same random bs issues with these 2 right?

[simple ore]

gpt sovits tts

I can help with Applio issues if there are any, for RVC good luck

[royal juniper]

I wanna use it to train voice and mix different peoples voices. it can do it right?

[simple ore]

yes

[royal juniper]

ok thank you. lemme check the program

[royal juniper]

applio0 has more functionality?

damn lemme try

pth files work on it right?

[simple ore]

yes, same inputs

it has some extras, but basics are the same

[dark walrus]

1303446885423644735

[flint aspen]

how do i mute the rvc from previewing my voice everytime i speak

[lavish trout]

Hi can you guys show me how to make an AI cover with a private model?

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[pastel oak]

You can check this out

[low shard]

You're using an outdated colab,
What are you looking for and what's ur pc gpu?

[ember panther]

hey is there any model training site you would reccomend doing it on my computer seems taking so long

[low shard]

yea ofc there are cloud ways, btw its normal it takes hours

What's ur pc gpu?

[lavish trout]

Hey so for some reason my huggingface model doesnt have the index file

It only has Readme and gitattributes. What do I do?

[ember panther]

rtx 3060 laptop gpu

[low shard]

Are you uploading ur model or found a random one

[low shard]

Memory GPU?

[ember panther]

wdym by memory gpu

[rare goblet]

Ayo? @ember panther level 3 !!!

[low shard]

The GPU memory

[ember panther]

must be 6gb

[low shard]

U can check it here

In task manager, via Ctrl shift eac

[ember panther]

7.9 gb

oh 13.9 gb

[low shard]

I mean the GPU memory, not the shared GPU memory

[ember panther]

gpu memory is 13.9 shared is 7.9

dedicated is 6

[low shard]

Yeah then it's 6

Kinda low

And yes laptop GPUs are weaker than normal ones

Technically you could do it locally

If you really want to do it on cloud (remote good pc):

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

Not really sure what would be the best in ur case

[ember panther]

im trying to make an olivia rodrigo model since she is so popular is there a lot of models of her already?

i mean i searched finded a lot

but the sound was kinda robotic

[ember panther]

which one would you suggest for the quality

or all have the same quality?

[low shard]

yea it's the same program

[ember panther]

alr thanks have a good day!

[knotty moth]

3060 desktop is more recommended for model training, but 6 GB is still fine for batch size 4 (recommended for less than 5-10 mins dataset)

[stray steeple]

Hey does anyone know how to make a voice sound real, I got the vocals to the song I wanted to make ava max sing but it's bad sounding. I used Ilaria_RVC, would really appreciate some guidance. ty! (EDIT: Looking up in chat too)

[zenith fog]

what about 3050 4gb?

[brittle wing]

hey, ive been tryna import models into the rvc and whenever i try switching to them it doesnt let me

it gives me an error message that says "ERROR: Exception in ASGI application"

any fix??

[wintry forge]

hello

[rare goblet]

Ayo? @wintry forge level 1 !!!

[wintry forge]

anybody here?

I downloaded some sounds I tried to use in voice ai but it says it haves to be mp3. or something like that

what program do you use?

please help

[pastel oak]

Do you want to do AI covers or use realtime voice changing

[brittle wing]

Hello, new here. I wanted some information regarding the voices in "voice-models".
I am trying to create a project for a youtube channel, and I need a female voice. I came across the AI RVC and I loved it!
My doubt is: Can I use the voice models for Youtube monetization? Basically, the video contains mostly my voice, but also the female one. It's a complex project that I have in mind. With long duration.

I read most rules, but I didn't read anything regarding something like this. Anyone can clarify? I don't want to cause trouble. Nor anything related to copyright strike.

[lavish trout]

Yes its mine

[low shard]

Yea, YOU have to upload the .pth and added index of ur model

[lavish trout]

How do I do that?

[low shard]

read https://docs.ai-hub.wtf/essentials/voice-models/#uploading-to-hugging-face

[lavish trout]

Is there a way to do this on mobile/browser

[low shard]

Yes, its explained how to upload it

did u make the model first of all?

[lavish trout]

Yes I used rvc disconnected

Before I ran the cells, I made the zip file with the audio dataset I wanted to use. When I finished the rvc colab, there was a new folder that popped up with the name of my model

The folder has config.jason, events.out.tfevents and rvclogs.zip

[lavish trout]

Hey what do you think of what I just told you?

[rare goblet]

Ayo? @lavish trout level 2 !!!

[low shard]

u should have also a

.pth and added index

[knotty moth]

you should have the voice owner consent with you

[brittle wing]

Hello, Can anyone please give me the link where I can download the voice changer cuz I forgot

[low shard]

realtime voice changer for calls? What’s ur pc gpu?

[lavish trout]

It doesnt have that in the folder.

[low shard]

did u train the model and the index?

bc it should be in the model folder, in rvcdisconnected folder, in drive

be sure ur not following yt tuts

[lavish trout]

I was... I restarted and deleted everything and will now be following the tut you sent

[lavish trout]

Im not sure I see where it says I can do it on mobile in the link you sent

[low shard]

if u don’t have the .pth and added index u can’t upload it

U will need to train it

i guess u got disconnected mid training?

[lavish trout]

I know but I mean restarting and making the model correctly now

[low shard]

also never follow yt tuts, those are all the cloud(remote good pc) ways for training:

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

[low shard]

the tutorial i sent before its only for uploading, now i sent u all the cloud ways to train

And yes u can do it on mobile, it will be just not comfortable and kinda a pain for the interface, but the program runs on a good remote pc

[lavish trout]

How do I make a folder into a zip file on google drive mobile?

[low shard]

not the whole folder, u should download the .pth and added index, zip it in your phone file manager (it depends by app and phone, u could just google it), then upload it

but u said u don’t have the .pth and added index

[lavish trout]

Ahhh so I download the pth and index

I thought the program makes it for me when I train the models. So I download it before I train?

[lavish trout]

Youre right I dont have it

[low shard]

.. that means u didn’t train the model

[low shard]

you can use any of those ways to train

Google colab = ez but low gou time
Kaggle = harder but way more gpu time

[lavish trout]

I'll use google colab

[low shard]

when it trains the model, it makes those

but it seems like u didn’t train it correctly

which is why u should never follow yt tuts for this

[lavish trout]

I didnt

[lavish trout]

I know now

[lavish trout]

I just downloaded the D and G pth files

From the how to use RVC mainline page

[low shard]

Also when u train u have to use the tensorboard, https://docs.ai-hub.wtf/rvc/resources/epochs-tensorboard/, in training

I suggest more kaggke as u have less risk of being disconnected but its harder

[low shard]

Those are used for pretrains.. u cant use the model like thqt

the .pth should be like modelname.pth

[lavish trout]

I downloaded the tensorboard

[lavish trout]

So I should rename the folder?

I dont understand

[low shard]

u won’t need to, its said that u don’t need to download it for cloud like google colab, it will be already downloaded

[lavish trout]

I see

[low shard]

No.. can you just show me the WHOLE model google drive folder?

a screenshot

[lavish trout]

I currently only have the folder with the name of the model I want to make with the audio dataset I have inside of the folder

[low shard]

that seems to be just a dataset, which is even bad to use a .mp3

you didn’t train anything

[lavish trout]

I said I wanted to make a new AI model and start over, this is me starting over so I can do things correctly. Now Im wondering where to go from here

[low shard]

I meant a screenshot of the previous folder in case u missed the .pth

the one that u seem to,have trained

[lavish trout]

I deleted it

Since you said it wasnt correct, so now Im starting over to do it correctly

[lavish trout]

It was originally an m4a file and then I changed it to an mp3 file. Its for personal use so its not like it has to be top notch quality. I'm just trying to quickly make something, and wven managed to make a voice model on Speechify the website, but it wouldnt let me have the model voice over an audio file, and would only let me do text to speech.

[rare goblet]

Ayo? @lavish trout level 3 !!!

[low shard]

speechify is shit, just paywalled rvc, the only site that lets u make a model for free daily is https://weights.gg , that still uses rvc, and ofc also the cloud computing ways

the dataset quality matters alot btw, and its gonna take hours to train

it wont be smt that quick

[lavish trout]

I know, but like figuring this out has taken a couple days when I assumed it'd atleast take one day for a few hours

[low shard]

So:

1. Would need to check https://docs.ai-hub.wtf/rvc/resources/datasets/ & https://docs.ai-hub.wtf/rvc/resources/vocal-isolation/ for adjusting your dataset, if u have a low quality dataset it’s gonna be bad
2. Use any of #✨│ai-help message which almost each one of those has a guide

[low shard]

yea its bc u wasted alot of time on yt tuts

[lavish trout]

Will my model be on a public database or can it be private use? Also I'm not sure where yo find the create model option

[low shard]

its on the top https://www.weights.gg/it/train/voice, and the privacy depends on what u choose

tbh i never used weights.gg for training rvc, we are partnered with them

[brittle wing]

what does this mean "* Line 1, Column 1
Syntax error: value, object or array expected."

[rare goblet]

Ayo? @brittle wing level 1 !!!

[low shard]

What are u using and doing?

be sure to not follow random yt tuts

[brittle wing]

used an old link that worked before from a friend

[low shard]

which? Send it

and also tell me what ur trying to do

[brittle wing]

https://huggingface.co/wok000/vcclient000/tree/main

[low shard]

oh wokada for realtime voice changing for calls

what’s ur pc gpu

[lavish trout]

Thanks, this is exactly what I needed!!!

[brittle wing]

rtx 3060 12gb

[flint solar]

its js like kits nun too crazy

[low shard]

i personalky used to do it manually with google colabs, but yw

[low shard]

hell nah not like that 😭

kits.ai asking 60 dollars a month

weights.gg is free

[flint solar]

can u export the model?

[flint solar]

WHAT

[low shard]

and kits didn’t care about quality (like if the models had an added index)

[knotty moth]

model training is actually costly, your best bet is to use a colab/kaggle notebook unless you have decent GPU

[low shard]

yes ofc u can download any models by the 3 dots and download button

[flint solar]

bro i remember it was free

😭

[low shard]

like year ago or more that i wasnt even staff

now they are completely different

they aim for royalty free voices or smt, kinda different

and asks 60 dollars a month for 12 custom voices

its still rvc tho

[flint solar]

itrain on kaggle

[low shard]

its free on weights.gg too btw

[flint solar]

dats so dumb

[low shard]

don’t ask me how much money they have

but its free lol

I don’t use it for rvc models, but for Flux.1-dev LoRAs

[flint solar]

weights is acc kinda cool

u can train loras too

[low shard]

yea for free

They have also characters

But tbh i mostly have fun with images & loras

[flint solar]

its good

[low shard]

yea

[low shard]

my bad i forgot lol
Well, technically that should work by following https://rentry.co/VoiceChangerGuide
Even if id suggest u to use the otimized fork instead https://rentry.co/VoiceChangerGuide

However im not a wokada helper, if following none of those guides work for you, its better u ask in #🔍│help-w-okada

[brisk void]

Hi! I am tryng to mix two voices with Applio, however I am getting a Value error:

Traceback (most recent call last):
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\gradio\queueing.py", line 536, in process_events
    response = await route_utils.call_process_api(
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\gradio\route_utils.py", line 321, in call_process_api
    output = await app.get_blocks().process_api(
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\gradio\blocks.py", line 1935, in process_api
    result = await self.call_function(
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\gradio\blocks.py", line 1520, in call_function
    prediction = await anyio.to_thread.run_sync(  # type: ignore
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\anyio\to_thread.py", line 56, in run_sync
    return await get_async_backend().run_sync_in_worker_thread(
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\anyio\_backends\_asyncio.py", line 2441, in run_sync_in_worker_thread
    return await future
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\anyio\_backends\_asyncio.py", line 943, in run
    result = context.run(func, *args)
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\env\lib\site-packages\gradio\utils.py", line 826, in wrapper
    response = f(*args, **kwargs)
  File "C:\Users\user\Applio\Applio-3.2.6\Applio-3.2.6\core.py", line 618, in run_model_blender_script
    message, model_blended = model_blender(model_name, pth_path_1, pth_path_2, ratio)
ValueError: too many values to unpack (expected 2)

Does anyone know why this may happen? Both pth files are voices that work fine on their own, and the relation is 60-40 (it fails with 50-50 too)

[simple ore]

likely trying to merge two models with different sample rates

[brisk void]

It may be the case. I trained one of them (48k) but the other one is downloaded, how can I check its rate?

[simple ore]

model info

[brisk void]

Maybe is a bit of a silly question but...where can I check that? I just have the pth and index files and I do not see any model info option on Applio GUI

[brisk void]

Thanks mate!

I tried with 2 48k models now and this is the different error I get:

TypeError: cannot unpack non-iterable EOFError object

[simple ore]

just unzip two models manually

[wicked tinsel]

does anyone have a tutorial or can explain the basics of the settings to me? My voice models always sound totally robotic even with good models with normally very good results. I use RVC_HFv2

[remote steeple]

help pls

uhh

cmd stuck on

loading web thing

i-dk

127.2o9529528952952 9592 925

something like that

and the app is white blank

what do i do

pls helpp

nothing load

heeeeeeeeeeeeeeeeelp

[simple ore]

post it 50 more times

that'll help

[remote steeple]

pls help bro...

[wicked tinsel]

where can I upload voice models in rvc_HFv2 I only see how to add them via url under recources

[low shard]

he's gonna do it fr 😭

[low shard]

Are you using https://huggingface.co/spaces/r3gm/RVC_HFv2 ? It's old, it runs on CPU

Use Ilaria RVC Zero instead, runs on ZeroGPU, A100 GPU, meaning its way faster

[wicked tinsel]

thank you so much

[rare goblet]

Ayo? @wicked tinsel level 1 !!!

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[low shard]

read the ilaria rvc zero guide

u can manually upload models here

[wicked tinsel]

thankssss

[low shard]

or upload by link

yw

[remote steeple]

can someone help

me with rvc

[azure marsh]

Hey, b! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[low shard]

What's the issue?

[remote steeple]

i start http thing

cmd show up

uhh

3 second later

some app show up

but it shows nothing

it look like this

then theres some

[low shard]

oh that's wokada not the rvc program, im guessing ur doing realtime voice changing

what's ur pc pgu?

[remote steeple]

pgu

oh psu

uhh

700w

[low shard]

that seems to be a charger ?

[remote steeple]

you men

mean

gpu

[low shard]

yes

[remote steeple]

4060 ti

soon im getting

4070

super

it stuck with this

[low shard]

oh its good yea,
I'm not a wokada helper,
But i can tell you that

• for the Original wokada you could try checking https://rentry.co/VoiceChangerGuide
• And for the fork (optimized, better performance) wokada u could check https://rentry.co/ForkVoiceChangerGuide
  Just in case u followed some random youtube tut

If you still have issue with the guides i gave you, it's better u ask in #🔍│help-w-okada

[remote steeple]

how i instaall

fork

[low shard]

by reading the guide

[remote steeple]

i have nvidia

i instal

amd or nvidia

i have amd

cpu

amd

[low shard]

btw, i just remembered a wokada helper saying that deleting stored_settings.json and restarting could make it work, maybe try that and restart

[remote steeple]

okay

[low shard]

u got an rtx 4060, it's made by Nvidia

[remote steeple]

my cpu amd

ryzen thing

amd ryzen

[low shard]

yea but u should choose based on ur gpu

[remote steeple]

okay

[low shard]

anyways, either try this or use the fork

[remote steeple]

this

IT FIX

[low shard]

i think so

Alright

[remote steeple]

wtf

my settings

brooooooooo

wait no it back

[simple hound]

what is config.json?

i dont have it in voice models

[pastel oak]

I can tell you are using an outdated version of wokada just from that comment alone

pls use a more up to date version from the guide

-rt

[azure marsh]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[pastel oak]

if this has nothin to do with wokada then gg ignore me

[rare goblet]

Ayo? @simple hound level 1 !!!

[quasi dagger]

Which RVC are you using? Also it says 'no module named numpy' which means it didn't install correctly

[simple hound]

im done, ill try this later

[old kiln]

is there better online converter than that https://huggingface.co/spaces/TheStinger/Ilaria_RVC

[rare goblet]

Ayo? @old kiln level 3 !!!

[old kiln]

because llaria rvc is little to robotic in some higer pitches can i fix that?

[marsh cradle]

https://huggingface.co/spaces/IAHispano/Applio

[old kiln]

is really better?

[brittle wing]

What will happen if I include backing vocals in a dataset?

[marsh cradle]

That depends on each person, although it is true that more than 1 million people use it. I recommend you try it and draw your own conclusions.

[brittle wing]

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[empty notch]

Can anyone help me why the file won't load?

[azure marsh]

Hey, !! Please use the command !howtoask to increase your chance of getting help by structuring your question in a way others can understand better. Also make sure you're asking in the right help channel:

• General RVC help: #✨│ai-help
• W-Okada / Realtime RVC: #🔍│help-w-okada
• AI image related: #🔍│help-ai-art

[empty notch]

!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.

[brittle wing]

How so many steps at 70 epochs already

[wintry forge]

realtime voice changing I want to troll my friends

[pastel oak]

-rt

[azure marsh]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[pastel oak]

First link is optizimited version, 2nd original

[lavish trout]

Are my creations on weights private if its a private model?

How do I make my creations private and not public

[tawdry perch]

how do i fix my mic going quiet after using

[low shard]

im guessing u talking about wokada, its better u use #🔍│help-w-okada

[low shard]

if u are going to make one, u have this checkbox to make it private

If u already made it, go on the model page -> edit model -> change visibility -> private -> save

[lavish trout]

I made it private, but will covers using the model be privage

[low shard]

all covers u make are private unless u manually share it to a friend

[lavish trout]

Got it

[turbid wolf]

is there any some sort of something that i can upload more than 1 file audio like rvc zero ? im trying to make a l4d2 mod that has over 2k sound files and rvc zero keeps disabling my progress after a bit and puts a 2 and half a hour timer

and i cant seem to find any other colab or huggingspace that lets me upload more than 1 files too..

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[flat onyx]

Is there any free alternative besides Google Colab for making AI Covers?

(Because I could only run AICoverGen using Google Colab for 3 hours)

CPU: Ryzen 5 3500U

[marsh cradle]

https://docs.applio.org/getting-started/other-alternatives#alternative-3-kaggle

[brittle wing]

Is it overtraining

[azure marsh]

Overtraining: un modo efficace per rilevare il sovrallenamento è controllare se il TensorBoard Graph inizia a salire e non torna mai più giù, portando a un output robotico e ovattato con scarsa articolazione.

[brittle wing]

Should I untick 'ignore outliers in chart scaling'?

Unticked and still doesn't look like overtrain

[low shard]

Btw ur cpu means nothing here, it's running on cloud (remote good pc) not locally (ur pc), What I would need to know is your GPU

Im guessing u got a bad pc (It's better if you tell me your pc gpu first)

Anyways, for cloud alternatives there's:

• Ilaria RVC Zero fastest for free on cloud
• Applio Kaggle 30 hours of better gpu than colab, weekly for free
• Use an alt google acc

Or if you got a good pc gpu, do it locally

[low shard]

the lowest point seems to be around 10k steps,
U could try to let it train a little more, if it doesn't go any down then it's overtraining

[brittle wing]

Huh I used titan pretrain and seems like it finished at 200 also it kept going down it's only gonna mess up

[grave star]

i have 1080oc (gigabyte), could someone who also have similar gpu tell me optimal settings?

[flat onyx]

Oops, yeah my bad. So my GPU is Radeon Vega 8 Graphics.

Regarding Ilaria RVC Zero, I tried converting an entire song (background music + original vocals), but the output only consisted of RVC vocal (without the background music)

[low shard]

yea that gpu is not good enough for local

Regarding Ilaria RVC Zero, I tried converting an entire song (background music + original vocals), but the output only consisted of RVC vocal (without the background music)
Yes rvc works only for vocals, only aicovergen automatically separated instrumentals and inferenced vocals, for every other rvc forks you have to do that manually

I think the only other cloud ways that separates it automatically is https://weights.gg

bc rvc is not used only for ai covers

[low shard]

check https://docs.ai-hub.wtf/rvc/resources/vocal-isolation/ , you have to do the cloud ways to separate vocals and instrumentals

[flat onyx]

I also found that AICoverGen is on HuggingFace, but I encountered "'NoneType' object has no attribute 'setdefault'" error when converting

[rare goblet]

Ayo? @flat onyx level 1 !!!

[low shard]

are you talking about https://huggingface.co/spaces/r3gm/AICoverGen ?

[flat onyx]

yep

[low shard]

i remember that one is broken

u can't use it

[low shard]

your only alternatives are either do it manually, or use weights.gg that does it automatically

There is no AICovergen for Kaggle nor HuggingFace spaces that work

[flat onyx]

Ahh so if I use weights.gg, I don't have to use Ilaria RVC right?

[low shard]

Its your choice on what to use

they all just use RVC, the really lazy and easiest ways are either weights.gg that does it automatically

or ilaria rvc zero that u have to separate the instrumentals and music manually, but also its pretty easy

Choose whatever fits you better, both are free

[flat onyx]

Thank you very much!

[low shard]

Your welcome!

[quasi dagger]

Yes, it's recommended to click on the ignore outliers

[brittle wing]

Huh is that why stuff always looked overtrained to me???

Does it mean ...

I still have to train

[low shard]

did u pheraps mean checked like this in the pic

bc that should always be checked on

[brittle wing]

SIIII

Voglio dire yes

[quasi dagger]

Yes

[low shard]

yes u should always leave it checked on

[brittle wing]

Ticked or unticked

The guide says to untick

[low shard]

it says ticked, to activate it

[brittle wing]

Allora il modello e finito a trainarsi

[low shard]

btw he meant to de-activate it, untick it, while in the guide it says to tick it
I think u misunderstood him

[low shard]

devi vedere come va il training

[brittle wing]

?la devo attivare o no

[low shard]

non significa che devi fermare il training

pero devi lo stesso vedere l'andamento del training

[brittle wing]

Tick or untick

[quasi dagger]

Sorry for the confusion, I mean click on it to activate

[low shard]

tick

lets speak english

its better

[brittle wing]

Then it has finished training my model is ready

[low shard]

can u show me a screenshot of the graphs

[brittle wing]

@low shard It's the latest screenshots I sent

[quasi dagger]

Around 9k steps seems good

[brittle wing]

It's already finished completely cause I used Titan

It has less epochs but a lot of steps.
Aren't steps more important

Just like SVC the steps mattered back then

Not epochs

Somebody told me that steps matter more

[quasi dagger]

That's normal, the steps are always more than the epochs

It's based on the dataset length, if you have a longer dataset it will show more steps for the same epoch

[simple ore]

lets say you have 1000 equal audio slices.. you run them with batch 4, you get 1 epoch consisting of 250 steps.... you run them with batch 8 you get 1 epoch consisting of 125 slices

batch size is how many steps the application is trying to evaluate in parallel

[brittle wing]

I'm lazy to calculate rn

[rare goblet]

Ayo? @brittle wing level 13 !!!

[brittle wing]

The model is finished

[simple ore]

batch 4 - a lot of small adjustments per epoch, batch 8 - half as many adjustments, but each is bigger

[wintry grotto]

do yall know if theres a good way to remove background vocals?

[simple ore]

using a pretrain is like using a pre-baked cake

[quasi dagger]

Yes, you can use https://mvsep.com/ and choose a karaoke model which separates the lead and background vocals.

You can reduce the queue there if you create an account

MelBand is a newer model so I recommend you go with it

[wintry grotto]

i’ll try it when i’m back i’m my computer

[gloomy pike]

guys help, me cmd saying Pipeline is not initalized

[pastel oak]

Install failed, delete pretrain and model_dir and then relaunch the program

[gloomy pike]

Decryption_failed_or_bad_record_mac

[visual fiber]

guys... where's the voice model?

[hard juniper]

are u guys having problems loading new samples?

becausa the only samples that are working with a good delay are the ones that came with the instalation

[pastel oak]

If you have an AMD GPU using the original wokada directML, you need to export your new uploaded models to onnx

[low shard]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models :
  • Send " @gusty kestrel search (name of the model)", without the ()
  • Do /find with @earnest musk
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

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

[ripe jacinth]

Is there like an RVC api? I'd like to use RVC programmatically

[marsh cradle]

you can use this repository, it is not an api and it runs locally on your pc but you can create great programs using RVC_CLI as a backend

https://github.com/blaisewf/rvc-cli

[hard juniper]

and how do I do that?

[pastel oak]

There is a button next to save setting called "export to onnx"

Or you consider changing to fork wokada which runs better for amd and doesnt have this onnx problem

https://rentry.co/forkvoicechangerguide

[hard juniper]

tks

[hard juniper]

and what do I do after extracting?

[pastel oak]

After export to onnx, you reupload that onnx file as model slot

[brittle wing]

Hey guys! What's the best Pre Trainer for Female Singers atm?

[pastel oak]

original

[brittle wing]

Why's that?

[quasi dagger]

People found out that all custom pretrains have some sort of problems like duplicate harmonics or weird lines showing in the audio spectrogram...Now it's recommended to use the original pretrains

[final tundra]

how to fix this

all time fcking windows delete the files

[quasi dagger]

Did you install RVC on system32 folder? Try on another location

[steel girder]

Hii

Can someone help me? How do I create an RVC voice model? Or any link to a collab for training please

[low shard]

What's your PC GPU?

[marsh cradle]

https://docs.applio.org/applio/beginners/first-model

[silent stratus]

if i have a 5 min dataset should i use ovo 2 super or default pretrain

[steel girder]

sorry for just replying now, I use phone

[low shard]

oh yea u can't do it locally (on ur device), u have to use cloud (remote good pc)

As you dont got a good PC, its better you use cloud (remote good pc) for training an RVC Voice Model:

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

Google Colab = Ez but low gpu time
Kaggle = Harder but more gpu time

[gleaming kindle]

its taking 2 minutes on average to train a single epoch, is that normal or am i doing something wrong

[steel girder]

I'm sorry but I didn't understand anything, I just wanted a link to train the model because I remember that in April of this year I trained voice models on my cell phone using a Google Collab link but it doesn't work anymore

Let's say I'm a beginner, even though I know how to do things to train, I just wanted the training link

[low shard]

Yeah there's way more ways now

I told you all possible cloud ways

There's 3 main cloud computing services, colab, Kaggle and lightning.ai

[steel girder]

can it be the collab?

[rare goblet]

Ayo? @steel girder level 1 !!!

[low shard]

The easiest ones are Colabs but the GPU time is not granted, you have the right of getting disconnected

[steel girder]

but they are not models of very long time, so no problem

[low shard]

The best one in terms of performance is Kaggle as it gives 30 hours weekly of better GPU, it's gonna be harder than a colab

[low shard]

Training a model takes hours

There isn't a right amount of epochs, so u have to use the tensorboard

[steel girder]

it's because I don't know how to use this kaggle

[low shard]

Yea that's why I sent u guides

I'd personally suggest to use kaggle, if it's too hard for u IG u could use colab

[steel girder]

aaah now I understand I think

but where can I get the website link? I can't find it in the guide

[low shard]

for which one?

[steel girder]

is this it?

It's because nothing is showing up to train

[low shard]

that's the applio colab yea

[low shard]

you have to read the guide for applio colab https://docs.ai-hub.wtf/rvc/cloud/applio-colab/

[steel girder]

AAA OK OK

Thank you so much

I am going to try

[low shard]

alright yw

[steel girder]

Tysmm

[steel girder]

I did exactly what the guide said

I even put the dataset link, which wasn't even asked for in the guide, but it still didn't work

[rare goblet]

Ayo? @steel girder level 2 !!!

[brittle wing]

I haven’t done original pre trainers for a while.

However, I have noticed realistic dynamic changes with vocals.

The transitioning from quiet singing into full range is a huge difference from my experience with pretrainers like KLM.

Maybe you’re right, I’ll give original a try.

Edit: I heard KLM 4.2 is cancelled unfortunately.

[knotty moth]

KLM 4.2 suffers that said issue, you prob mean 4.1?

[brittle wing]

No, I heard the creator for KLM 4.2 stopped working on it.

I personally think 4.0 sounds better than 4.1

[knotty moth]

again because of that said issue

[brittle wing]

What do you mean?

[knotty moth]

read the previous comment before mine duh

[knotty moth]

I don't recall comparing with 4.0, but "better" in what aspect? pitch range (esp above A5), or else?

[lusty kindle]

uh

"ERROR: [Errno 10048] error while attempting to bind on address ('0.0.0.0', 18888): only one usage of each socket address "

[brittle wing]

No, it was a personal statement from my end.

Not what you said.

But, 4.0 has realistic dynamics than 4.1 and 4.1 has more background noise.

Edit: pitch’s about the same.

[lusty kindle]

it just

crashes

[knotty moth]

less sensitive to noise huh?
did you use the 32k sample rate? 4.1 doesn't have 40k option tho

[brittle wing]

No, I tested 48k on both with high quality datasets.

Edit: but for me, 4.0 seems to sound more like my original model when pronouncing words while singing.

[lusty kindle]

whats the best settings for quality

[brittle wing]

Depends

[lusty kindle]

how so

[latent yacht]

Help

I upgraded to a newer GPU but when I use the voice changer, it shows only cpu instead of the GPU for usage

I am unable to send a picture in this server for some reason

please ping me if you know how to fix this

[rare goblet]

Ayo? @latent yacht level 1 !!!

[latent yacht]

ok now i can send the screen shot

[thick mural]

Help

Im stuck on the white screen when trying to open the voice changer

I watched a video and they said it shouldnt take longer then a minute to load

[fading lodge]

what version do u have

@latent yacht

[latent yacht]

I have an AMD gpu btw

[fading lodge]

That's the right ver

[fading lodge]

Ohhh

That's why

[latent yacht]

can i be helped?

[latent yacht]

or am i cooked?

[fading lodge]

https://huggingface.co/wok000/vcclient000/blob/main/MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.18a.zip

Download this

U have onnxgph cuda

Cuda is nvidia

But onnxdirectML Cuda is for Nvidia cards

Try that and it should work

👍

[latent yacht]

@fading lodge if there was a more better word than THANK YOU, I would use that

[fading lodge]

No problem

Also wajt

When u open the voicechanger

Dm me ur settings

I'll help u set it up

[latent yacht]

a genuine helper? holy shit THANK YOU SO MUCH

I will

[fading lodge]

😭

I've been using Okada for a while

[rare goblet]

Ayo? @fading lodge level 8 !!!

[fading lodge]

I tend to help others

[sick sable]

Hey, what is "epochs" and how do I configure it in app?

Also where can I configurate certain settings, because in voice-models channel there was text preset for certain voice:
Dataset: 1:11
Sample Rate: 32k
Pretrain: Original
Pitch Extraction Algorithm : RVMPE
Batch size: 2
Hop Length: 64
Made on RvcDisconnected

Where do I need to change it in Voice Changer app?

[latent kettle]

Epoch is a training cycle. Just like if you read a book for 1 time, it's 1 Epoch

[sick sable]

Cool, where do I change it in voice changer app?

[rare goblet]

Ayo? @sick sable level 1 !!!

[latent kettle]

These settings are used when you are training a model

[latent kettle]

If you are training a model, then these settings are used

[sick sable]

So that the user understands how the model was made?

Aha, I just mentioned somebody making text for correct configuration so model runs best on settings author writed

[latent kettle]

Yes

If you want to train your own model, you can experiment with these settings to get a high quality model

[latent yacht]

@fading lodge hello?

@fading lodge I need you cuh

[sturdy stratus]

Gtx 1070ti Is it okay if I use it with voice AI?

[latent yacht]

i used it on Gtx 750ti, so don't worry about it

[low shard]

for realtime voice changing, wokada, it should be

be sure to use the wokada fork for better performance btw

[sturdy stratus]

Thx

[low shard]

-rt

[azure marsh]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[low shard]

1st link is the wokada fork

voice.ai sucks ofc

[sturdy stratus]

I want it for singing

[low shard]

so.. you want to make an ai cover, or use the voice in realtime voice changer for calls?

there isn’t just a single program for both of those so

[sturdy stratus]

Can it be used for chorus?

[rare goblet]

Ayo? @sturdy stratus level 3 !!!

[sturdy stratus]

I sing and have chorus in some parts of the song.

[low shard]

oh so record urself doing chorus and then change your voice to the ai one, ig u can do that yea

you will need RVC tho, wokada is used only if you are going to do it in realtime for calls

Your GPU is good enough to do inference (use models) locally (on ur pc), you won't be able to train (make models) but use them

You can:

• Locally (runs on your pc so the speed depends on that, you will have to set it up with the guides):
  • Applio: A fork of RVC with some extra features like Applio TTS, kinda faster and simpler but same quality tho
  • Mainline: The original RVC
• Cloud (remote good pc, easier and faster than ur PC but it's limited):
  • Ilaria RVC Zero: fastest and simplest that you can get for free
  • Weights.gg: 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

[sturdy stratus]

Thanks, but I still have a lot to learn.

[low shard]

your welcome

[steep tree]

When I use my RVC model, the cmd shows "AttributeError: 'NoneType' object has no attribute 'dtype'." How can I fix this?

I hope someone can help me

[steel girder]

what does this mean? an error occurred extracting the index

[marsh cradle]

Make sure you have selected your model correctly

[ripe jacinth]

Yeah I have no clue how I'd even go about integrating that into my app
and most rvc python wrappers don't really have all the pitch extraction methods

[marsh cradle]

you can read the docs, it explains it quite well https://rvc-cli.pages.dev/Usage/rvc

[low shard]

you put a link, not a path https://docs.ai-hub.wtf/rvc/cloud/applio-colab/#b-dataset-path

[marsh cradle]

At first it seems complicated with so many arguments and more, but believe me it is quite easy

[ripe jacinth]

I think I'll do it by doing a pip install git+https://github.com/blaisewf/rvc-cli.git
and then that my project can compile into a native exe with all of this mess
nvm the command doesn't even work

[marsh cradle]

It is a command line (which you can also use as a python library but oh well) https://rvc-cli.pages.dev/installation

clone, execute install.bat and run with env/python.exe cli.py

[ripe jacinth]

no
merging my project with another!?
it'll create too much of a mess

[marsh cradle]

You could use it as a submodule in git and then use it as another library

You also have the kits API with generic voice models

https://www.kits.ai/api

[final sail]

For gasping, laughter, crying, sneezing, coughing, struggling do I need them in my data for the model to be able to do them

[rare goblet]

Ayo? @final sail level 1 !!!

[fading lodge]

Um

You can't do that

Sorry to break it to you

The most you can do is make breathing sounds

Depending on what you sampled

U can also try doing gasping

Mabye sneezing

But they sound very bad

[steel girder]

and how do I put the path? I've already looked in the guide and I couldn't find where to find it

[quasi dagger]

Hello! I'm installing Applio on colab right now to double check this

[steel girder]

Guys, now I did it, I was doing something wrong but now I learned, thank you very much for the help, I love you ❤️

[quasi dagger]

Nice to hear, happy training!

[simple ore]

well.. .why you dont select a dataset?

[quasi dagger]

That dataset dropdown only works if you have the files in the applio datasets folder, they were trying to import from a different place

[simple ore]

then use create data type and upload

[quasi dagger]

That's what I thought, but they figured it out already

[final sail]

Why it can't do them if it is training on them?

[storm mica]

Is there a version of RVC with batch convert functionality?

[marsh cradle]

https://docs.applio.org/applio/getting-started/inference

you have a section for it

[severe swift]

can sb help pls

i cant hear myself

its not working for some reason

also keep getting this but i dont understand the language

[pale berry]

I need help my app keeps giving a blue screen on the PC error code: IRQL Not Less or Equal

[low shard]

my app
Which?

blue screen on the PC error code: IRQL Not Less or Equal
What's your PC GPU & VRAM?

Also be sure u aren't following any yt tuts

That error seems related to your hardware or maybe drivers

[pale berry]

RTX 4060,vcclient_win_std_2.0.69-beta.zip

What is the best version for me, I didn't understand absolutely anything related to you or the server, what do you mean?

I mean

vcclient_win_std_2.0.69-beta.zip or MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a.zip
what's the difference

[rare goblet]

Ayo? @pale berry level 1 !!!

[pale berry]

just one example

[low shard]

Ah, wokada,
It's better you follow this guide as it uses the wokada fork that has better performance https://rentry.co/ForkVoiceChangerGuide#download-nvidia

[pale berry]

ok thanks for the answer

[low shard]

Yw, if the error still persists, it could be related to your drivers, it's better u ask a wokada helper in #🔍│help-w-okada as I'm not one and this is the wrong channel lol

[pale berry]

haha sorry about that, thank you very much

[sonic wharf]

what's best program to use for AI covers (Local and not cloud)

[spring orchid]

applio

[sonic wharf]

ty

[rare goblet]

Ayo? @sonic wharf level 1 !!!

[fading lodge]

because

its not implemented to do that

the voicechanger

or

wokada is used for voice changing

basically, a VOICE CHANGER

not a cough changer

or yawn changer

it dosent have a feature like that yet

mabye in the future it will

[pastel oak]

They are considered noise for voice. Having too much of them will "merge" these into the clean vocals and make even speaking sound completely bad. You can have very little seconds of them included that might help but not to your satisfaction

[fading lodge]

whatever this guy said

[modern holly]

Hey guys

What is the best real time voice changer

Wanna make girl voice

[flint solar]

w-okada

for more help go to #🔍│help-w-okada

[terse siren]

hey guys im making a dataset and how do i make them, the guide i am reading wont let me view the guide on making a dataset with appolio

[brave garnet]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra

Advanced Settings

Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low

⠀

Made with love by Ilaria 💖

[sweet dock]

with nvidia should i still convert to onnx?

[rare goblet]

Ayo? @sweet dock level 1 !!!

[indigo rover]

Is there any guide to making your covers sound better? Often I end up with some rasp and stuff and often I see stuff online where its almost as if the original sung it.

[brittle wing]

-hf

[azure marsh]

🤗 Huggingface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by thestingerx Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces
• 🆕 FaceFusion UI, by Nick088 Huggingface Spaces

More commands: -audio, - colabs, -kaggle, -local, -rvc, /help

[brittle wing]

-space

-colab

[azure marsh]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[charred jewel]

How do you install RVC again? The only one I have is from like July of 2023. I can't really remember where to get it from anymore.

[charred jewel]

https://huggingface.co/MangioRVC/Mangio-RVC-Huggingface/tree/main

Is it this one?

[simple ore]

"Updated: over 1 year ago"

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

⠀
Mainline RVC
Original project, suggested for advanced users,
by the RVC-Project team.

Applio
Simplified, suggested for all, by the Applio team.

RVC Studio
Simplified, suggested for all, by SayanoAI.

Mangio-RVC
Simplified, may not be supported anymore, by Mangio621.

AICoverGen
Simple yet great way to make covers, by SociallyIneptWeeb.

Replay
From the greators of weights.gg, excellent product for everyone.
⠀

Made with love by Ilaria 💖

[rare flint]

RVC is not the real time one right? i want to convert an voice audio to another voice

[charred jewel]

Is it the real one tho

[simple ore]

Mangio is a very outdate fork of RVC

see info above

[charred jewel]

Oh, well that's the one I have installed on my computer rn lol

ok

[brave garnet]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra

Advanced Settings

Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low

⠀

Made with love by Ilaria 💖

[charred jewel]

Why is Applio a public archive?

[brave garnet]

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs
Version 18a cuda

Download for AMD GPUs
Version 18a directml

Download for Intel GPUs
Version 18a directml

Download for Mac
Version 17b Mac
⠀

Made with love by Ilaria 💖

[simple ore]

https://huggingface.co/IAHispano/Applio/blob/main/Compiled/Windows/ApplioV3.2.6.zip

those links are outdated af it seems

[charred jewel]

oh

[simple ore]

what's your GPU?

[charred jewel]

Is Applio better than Mangio

Haven't used anything except Mangio for over a year lol

[rare flint]

what's the sysreq for Applio?

[simple ore]

it is one of the most up to date forks of rvc

for training 8GB+ VRAM, Nvidia ideally 2000 series+

AMD 5700+

[rare flint]

what about just voice cloning

i just need to change my voice to another

[simple ore]

inference can probably do with 4 or 6

[charred jewel]

I have an RTX 3060 Ti

[rare flint]

so 1650Super works?

[charred jewel]

Idk how much Vram tho

[rare flint]

3060 is like 6GB VRAM iirc, so youre good

[charred jewel]

I have 8 gigs

[simple ore]

1650 super is 4GB?

[rare flint]

yea

[rare goblet]

Ayo? @rare flint level 2 !!!

[charred jewel]

I've trained a lot of models on this GPU before, but I didn't know if the system requirements had gotten more demanding.

[rare flint]

how does ILARIA works?

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[charred jewel]

No training

[rare flint]

i assume this one is online right? but what are their limitations?

[charred jewel]

only inference

[rare flint]

oh ok

just what i need

don't need to download anything i assume (other than voice models)?

[charred jewel]

Idk how to actually import the voice models

I just took a quick look at it

But I thinks everything else is online, except for the voice models and audio files (If you wanna do a cover).

[rare flint]

i just need to change my voice to another voice for a placeholder on my project

that's all

[charred jewel]

Then you will need to have both model and voice recording on your computer.

That's all

Oh wait

You might need to upload the voice model to google drive or smth

idk how to upload it

[rare flint]

no it's just normal uploading

[charred jewel]

alr

[old kiln]

how upload model here?

[jaunty star]

oh man I don't know what happened 😦 for some reason the voice quality of my fork okada has been real bad lately as it started to crackle and stutter... I don't believe any setting has been changed during the last time I used it

the voice quality was real nice a month ago, same settings, much cleaner with no errors

[knotty moth]

check this guide https://rentry.co/voicechangerguide#crackle-fixes-windows

[thick aurora]

can't launch gradio on colab, the cmd shows AttributeError: module 'datetime' has no attribute 'UTC'

how can i fix this?

[placid wren]

I'm brand new to this stuff, so bear with me. I'm using Alltalk with SillyTavern, and I am experimenting with RVC. However, something is unclear to me. When I download a model from #1175430844685484042, I set it as the "RVC Voice" in Alltalk, but what do I set the "Character Voice" as? Do I use a high quality sample of the character? Or am I supposed to download some random high quality voice as the base, and RVC will make it sound like the character? If the latter, how do I know what "base" voice to use that will fit the character?

[low shard]

don’t use the public huggingface space, it’s public and running on cpu

im guessing u want inference (use model) on pre-recorded audios

What’s ur pc gpu?

[low shard]

It’s a fork of RVC, the major difference is TTS and the interface, for rvc check: https://gudgud96.github.io/2024/09/26/annotated-rvc/

[low shard]

Yes it’s online, the limitation is that your account got a zerogpu quota just like for every other zerogpu spaces,
Its the fastest but ofc not unlimited

Meaning u can’t like do 30 inferences all at once, the quota charges with time

amd u cant check the quota

check the guide where it explains more

[low shard]

https://docs.ai-hub.wtf/rvc/resources/datasets/

[final sail]

isn't all of that a part of voice with rvc too?

[final sail]

Why they are considered as noice when they are a part of speech?

[modern osprey]

Hello everyone, does anybody have a voice model of everglow?

[low shard]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models :
  • Send " @gusty kestrel search (name of the model)", without the ()
  • Do /find with @earnest musk
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

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

[pastel oak]

the frequencies a voice emits when speaking are not the same as crying or coughing, i dont have the scientifical terms and explanations for you

[cunning crag]

i have an issue that after i download something from applio the sound is only hearable in my left headphone

[modern osprey]

i can't find it

[low shard]

Read the 2nd part, u can make it urself or request someone to do it for u

[pastel oak]

open the weights.gg link i found plenty of everglow models in 2 seconds

assuming its the right one

[lavish trout]

Is there a way to change the picture I used for my weights.gg model?

[low shard]

edit model -> image -> upload or generate one

[lavish trout]

Got it ty

[low shard]

Yw

[solemn crystal]

what is "chunk" and "extra" work?

[rare goblet]

Ayo? @solemn crystal level 1 !!!

[low shard]

wrong help channel

For wokada use #🔍│help-w-okada

im not a wokada helper, but guessing u have the wokada fork u can see https://rentry.co/ForkVoiceChangerGuide#best-settings

[solemn crystal]

mb

[low shard]

or for the normal wokada https://rentry.co/VoiceChangerGuide#gpu-chart-for-known-working-chunkextra-on-the-official-okada-prebuilt-binaries-that-are-linked-above

[potent root]

what files do you need so you can continue training a model on a diff google account

[elder plume]

Does anyone here know of any RVC program that detects AMD GPUs

as i have been finding it very difficult to get one that works properly with them

@low shard you seem to be the mod here right? may i ask for some assistance?

[low shard]

Sure

[low shard]

Yep,
Just to reassure you are looking for the right program:

• RVC: training (making) models and inference (use models) on pre-recorded audios
• Wokada: inference realtime for calls

[elder plume]

yes, im finding a RVC training program

[rare goblet]

Ayo? @elder plume level 1 !!!

[low shard]

You're on windows right?

[elder plume]

yes windows 10

[low shard]

Good, you can use Applio (an RVC, so modified version, that is a bit faster and has a different interface with also TTS): https://docs.applio.org/applio/getting-started/installation#amd-gpu-support-windows

May I ask what's your GPU tho?

I don't know much about AMD ones but just to check it's not an integrated one

[elder plume]

im using a Radeon rx 7800 XT

CPU is ryzen 5 7600x

[low shard]

You're all fine then, by following the guide I sent you should be able to run it without issues

[elder plume]

well this is much appreciated

as i been tirelessly searching through yt

and they all seem either outdated or only for nvidia

[low shard]

Nah never search on yt for RVC, as the hype for ai covers kinda died, 99% of yt tuts are outdated

Written guides are better for being easily modified so more updated

[elder plume]

so for future reference, should i come here look for info or is there a place where ai enthusiasts go to?

[low shard]

For everything RVC-related (RVC, Wokada, forks of those programs, UVR) you can ask in the specific help channels here

[elder plume]

theres a problem with the link

[low shard]

Oops typo, fixed it

[elder plume]

okay, lemme try now

okay, so it works and it should give me instructions right on how to download?

okay, i understand the instructions now, thank you again

[low shard]

You're welcome!

[elder plume]

@low shard , how do you this particular step?

as i dont understand how to run command line from applio

[rare goblet]

Ayo? @elder plume level 2 !!!

[elder plume]

ive already unpacked the zip files

[low shard]

You need to go into the Applio folder, then write "cmd" at the top of file explorer where it tells u where u are, and it will open a Command Prompt (aka CMD) window, from there u can run the command

[elder plume]

okay, lemme take a look again

where do you find the cmd at the file explorer?

@low shard , this should be correct right?

now question is, which command line should i be using?

i ran the command that is given in that website

but got this result

[low shard]

I don't do it locally as i got a bad pc, but from your screenshot i seen you didn't download the pre-compiled applio

[elder plume]

oh, well now i feel like a dumbass

lemme go find it

[low shard]

Download this: https://huggingface.co/IAHispano/Applio/resolve/main/Compiled/Windows/ApplioV3.2.6.zip?download=true

then extract it

[elder plume]

so with this, i put into the library file

or do i leave the two separate

also is it necessary to have these files on a C drive

[low shard]

you neex to extract it in its own folder

[elder plume]

or that was just a suggestion on the website

[elder plume]

okay, got it

[low shard]

in the guide it's told to do that tbh

[elder plume]

okay, imma just move them over

i ask, since i wanna not download anything on C drive if possible

[sterile lichen]

Question, i got 36 audio samples should i merge them all to one audio before feeding it to model?

or does it not matter?

[pastel oak]

Dont merge them into one audio

[low shard]

doesn't that usually not matter as rvc should do auto splitting btw?

[elder plume]

does anyone know how to do this bit

as my gpu is listed as index 1

not 0

[pastel oak]

personally still had model collapsing with one file only vs. split up with labelsounds

[elder plume]

also i do not understand the 2nd step of adding this variable

@low shard im getting this after i ran the program

[pastel oak]

i mean it literally explains it, edit the file and change 0 to 1

[elder plume]

yeah, i found it now

but yeah im just very new to this stuff

[elder plume]

i genuinely dont understand why it aint working

[rare goblet]

Ayo? @elder plume level 3 !!!

[simple ore]

please follow the guide step by step

and read it

why you have libraries old when the guide says to check compatibity of your 7800xt?

[elder plume]

the problem is that, its telling me to add some things to a file which i dont see at all

[elder plume]

how do i check that?

[simple ore]

Download and install the VC++ Runtime.

done?

[elder plume]

yeah that part is done

its all set up

[simple ore]

Install HIP SDK

Check the System Requirements.

did you check requirements?

what did you do next?

[elder plume]

okay, let me take a look

so it seems 7800 uses gfx1100

[simple ore]

not that

https://rocm.docs.amd.com/projects/install-on-windows/en/develop/reference/system-requirements.html

[sterile lichen]

some files has only 3-4 sec of audio, will that be problem?

[simple ore]

7800xt -> green hip sdk

[elder plume]

oh

[simple ore]

okay, so you chose to download HIP SDK 5.7.1

[elder plume]

so its gfx1101

[simple ore]

what did you do after that?

[simple ore]

if the audios are already split and cleaned, dont use built-in slicer

[elder plume]

that one correct?

[simple ore]

``a. If your GPU has a green checkbox in the HIP SDK column:

Install either v6.1.2 or v5.7.1 HIP SDK from AMD ROCm Hub.

``

[pastel oak]

no problem , and do what noobies said, deactivate slicer in advanced

[simple ore]

and no process effects if using applio

[elder plume]

well it does have a green check box for my gpu

[sterile lichen]

where? i don't see it

[rare goblet]

Ayo? @sterile lichen level 1 !!!

[elder plume]

im real sorry for being smooth brain here, as i have no idea what the hell im doing

[simple ore]

downloaded a compiled version of applio and unzipped it where?

[simple ore]

why do you have library.old and rocmlibs?

did you do anything with ROCM in program files?

[elder plume]

no, i dont think so

i just ran the cmd on the applio file

and pasted this command line

[low shard]

mm weird

[simple ore]

after you press enter you get

paste 1st line, enter, wait until it is done

paste 2nd line, enter, wait until it is done, close

[elder plume]

oh okay, but what if i already repeatedly pasted both like 3 times

should i just delete all and unzip a new one

[simple ore]

then you installed pytorch fuck knows where

or did not at all

probably did not because there was no 'env\python'

[elder plume]

okay, so ill just do what you said about the lines

i was confused because they didnt specify you had to paste these lines in sequence

[simple ore]

after that

especially the last line

[elder plume]

am i on the right track?

[simple ore]

yes

[elder plume]

question is what do they mean by that file path

as i cant find it

[elder plume]

oh okay okay

[simple ore]

5.7 in your case

[elder plume]

the last ss there, how do i get that?

so i clicked open path

[simple ore]

double click on path, add new

[elder plume]

oh okay okay

[simple ore]

then browse

[simple ore]

yes

but 5.7

since you have 5.7 installed

did you install it? 😉

[elder plume]

yes i installed 5.7

the windows 10 and 11 one

so im on browse right now

what am i looking for?

[simple ore]

c:\program files\amd\rocm etc

same folder as you had, just 5.7 instead of 6.1

[elder plume]

okay hold on

[simple ore]

i sense 'installed' was "i did download the .exe' theme

[elder plume]

so i put it here right, cause that is the 5.7

im guessing

[rare goblet]

Ayo? @elder plume level 4 !!!

[simple ore]

@cursive agate explain what installing is

[elder plume]

as i accepted this and downloaded it, ran it

and thats it

[simple ore]

so if you ran it you should have a folder in program files

[elder plume]

okay, i didnt know there was a file created automatically

[simple ore]

what file?

when you run hip sdk .exe

regardless

the guide says Add C:\Program Files\AMD\ROCm\5.7\bin to your system's Path environment variable.

and conveniently providing you a value to add

[elder plume]

it leads me to this

[simple ore]

so after you click install

it installs it

[elder plume]

yeah its downloading now

[simple ore]

oh sweet ipad generation

they dont teach computer shit in schools anymore, do they?

[elder plume]

yeah like i said, i sincerely apologise for being very uneducated on how to handle these stuff

[elder plume]

well all i did for high school was a shit load of microsoft excel and access

a load of accounting work essentially

anyways its finished

[simple ore]

so.. did you 'install' VC Redist the same way?

[elder plume]

yeah installed it and ran it

[simple ore]

okay, so just add the path like I shown you above

[elder plume]

okay, lemme find it real quick

this is my program files

and i dont seem to see it

[simple ore]

AMD?

[elder plume]

that is for AMD adrenaline

when i first setup this pc

[simple ore]

go inside

[simple ore]

baby steps, I guess

go go

[elder plume]

okay

hey, its here apparently

so what next?