✨│ai-help

[latent sorrel]

nvm i should run STV

[signal cargo]

guys i'm currently using the ilaria rvc and unfortunately it takes too long until it suddenly crashes. What is the problem?

[proven hill]

huggingface?

[signal cargo]

ok it just worked miraculously... I will tell you if it doesn't work again.

[signal cargo]

and yes

[proven hill]

use the colab hun

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[grizzled forum]

how do i install tensorboard? this link doesn't seem to work https://us.aihubfrance.fr/guide-to-create-a-model/tensorboard-rvc

help what do i do

[brittle wing]

perchance could someone help me with the setup of this, ive got a fair bit done but im kinda lost

[fluid gyro]

Hey um.. idk how to start

The guides seem too pro

Didnt help... I just have the UI running but thats it idk how to proceed

[grizzled forum]

help i tried to run tensorboard but something broke

[fluid gyro]

oh nvm I think im going the right way

[odd shale]

Guides aren't that hard.

You're stuck on which part?

You can't train a model?

There's any error on your Paperspace console?

[fluid gyro]

sorry itsjust that Idk what is the order of things

and theres so many guides hahh

like hyperlinked

[red kayak]

okay make a folder and upload ur audio file in there

im assuming u know how to make a folder on paperspace

and upload

[fluid gyro]

yess, im doing the dataset thing now

[fluid gyro]

I wish it was as easy as that

Maybe it is tho

I know nothing

Im only an audio engineer but idk shit about aimodels

[red kayak]

its very easy

[fluid gyro]

really?

I have a really good dataset

clean af

[red kayak]

making a folder and uploading a file isnt that hard i would assume

[fluid gyro]

but like In the guide it says i have to pass it thru a colab or sum

and its very annoying

[grizzled forum]

is it ok if those are blank before i start training?

[brittle wing]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[red kayak]

nope

[grizzled forum]

they hadn't filled on their own

[red kayak]

u need to set up the venv

[red kayak]

u are missing pretrains then

[grizzled forum]

what should i do?

[fluid gyro]

sorry what

bruh im losing it with colab, it says in the guide that I need to run it there so that I can clone the "dataset"

[red kayak]

are u using colab or paperspace

[fluid gyro]

both seems like

[rare goblet]

Ayo? @fluid gyro level 3 !!!

[red kayak]

okay what is on ur screen atm

can u show me

[red kayak]

try switching back n forth

[fluid gyro]

dm?

[grizzled forum]

wdym by that

[latent sorrel]

who have the new version of colab?

[red kayak]

sure

[red kayak]

like click on 32k

and then back on 40l

[latent sorrel]

i have the outdated colab

[red kayak]

or whatever frequency u want to train on

[latent sorrel]

that why i have this issue

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

[red kayak]

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

[latent sorrel]

thank you

[rare goblet]

Ayo? @latent sorrel level 2 !!!

[rare goblet]

Ayo? @novel magnet level 2 !!!

[untold sierra]

how do i train an mp3 to a voice

[pastel oak]

https://docs.aihub.wtf/essentials/how-to-make-voice-models/

Everything explained in here very simple and direct

[untold sierra]

got it

thank you

im confused

how to i make an mp3 a pth?

[pastel oak]

You need to train the model like the 2nd step is saying

You choose either locally if you have a good gpu or choose on the cloud, then you need to set it all up and let it train

[untold sierra]

i clicked local, downloaded rvc, opened rvc, and now it just tells me to upload a pth

[rare goblet]

Ayo? @untold sierra level 1 !!!

[pastel oak]

There is a "train" tab

[untold sierra]

oh i see it

[pastel oak]

training a model is what you mean with "turn a mp3 into pth"

[untold sierra]

ok

thank you

another question

it asks me to copy the path of the training folder

im unsure where that is

and the picture that shows where "copy path" is doesn't show me where to find it

@pastel oak

[pastel oak]

manually put in your folder where your mp3 is in

[untold sierra]

my mp3 is in downloads

where do i put it from there?

[pastel oak]

then just select the downloads path file

C:\blablalba\bla\Downloads

you paste that in

[untold sierra]

i moved it from downloads into my ai folder. But where do i find the link?

the C:

ins't it usually there?

oh

i got it

C:\Users\Owner\Desktop\AI VOICE CHANGER

lol

[pastel oak]

Btw. dont use any spaces in your folder path

it doesnt recognize spaces

[untold sierra]

ok

[pastel oak]

so either change it to AI_Voice_Changer or just aivoicechanger

[untold sierra]

got it

[grizzled forum]

how can i continue training after restarting applio?

[rare goblet]

Ayo? @grizzled forum level 3 !!!

[low tangle]

if i trained for too many epochs (on rvc disconnected) how do i access the previous saves? do i put the step count here?

oops sorry senmou

[grizzled forum]

np

[pastel oak]

if you have this enabled, it should have saved a small version of each X epoch in the weights folder

So then go to the weights folder and check for the file

[pastel oak]

use the G and D paths to continue training

[low tangle]

in the mangio files?

[pastel oak]

It should be on here i think

[low tangle]

OH right

[pastel oak]

a folder called weights, maybe its inside another folder called assets

[low tangle]

thank you

[grizzled forum]

oh so i have to save them too...

i just exported my index and pth file...

[untold sierra]

@pastel oak sorry im almost done but i need help with this last step. My new voice is called "tek" and it isn't in the weights folder. I have the index file ready but not the pth

[rare goblet]

Ayo? @untold sierra level 2 !!!

[untold sierra]

this is all thats in my weights folder

@pastel oak ^

i believe those are just the default voices right? My new one isn't in there

[pastel oak]

Ye its the default ones

Idk did you disable the save final model option?

[untold sierra]

no?

[pastel oak]

I dont know what you did wrong then

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[ocean anchor]

is there any good

text to speech things

that are free

and i can use

voices like spongebob

[low shard]

Yup, you could either use any tts tools and then use the generated tts audio as an input for rvc using rvc models which are most ones in this server, or either use gpt so vits which is a tts ai that is better for those things but idk if theres a spongebob model already made for that

for everything you can check our docs https://docs.aihub.wtf

[ocean anchor]

ima just

not

because

idk wtf

any of those are

[pastel oak]

you scared him off with your choice of words damn

[low shard]

how could i explain it more easier

[pastel oak]

TTS tool is linked on that docs

Theres difference between TTS-made ai voices and the other ones that we call RVC

tts voices are called gpt so-vits. You can directly filter and search for those kinds on the server

[ocean anchor]

where

[low shard]

https://docs.aihub.wtf/

[pastel oak]

#1220108753240854680 message

this is the spongebob voice model @ocean anchor

[low shard]

wow didnt thought there was ona lmao

[ocean anchor]

what one should i use

[pastel oak]

I vividly remembered kimid posting this

[low shard]

gpt so vits has better quality

[ocean anchor]

what do i hit

[low shard]

u need to follow the gpt so vits guide inside the docs

its inside the tts folder

[brittle wing]

so ive got the audio ready and everything, can someone help me out with figuring out what to do now

[low shard]

you tryna make an rvc model or ai cover?

[brittle wing]

rvc model to use w a voice changer lol

[low shard]

this could help you out https://docs.aihub.wtf/essentials/how-to-make-voice-models/

if you dont understand smt or got some errors ask here

[pastel oak]

Quick insider info, at least from experience

your dataset should be 25+ minutes for realtime voice changer. Idk why but anything less might come out more robotic

[brittle wing]

ive got around that

[pastel oak]

guud

[sharp crystal]

Hello everyone, Could you help me please?

To import a model I must place it in /assets/weights and its log in /logs only, right?

I'm putting the address of my audio to the converter and I'm getting an error, it just won't work and I don't know what it could be.

Below would be the path of the audio I want to convert:
"D:\AI-Programs\pinokio\api\rvc.pinokio.git\app\audios\audio.wav"

Here is the audio model:
"D:\AI-Programs\pinokio\api\rvc.pinokio.git\app\assets\weights\SpongeBobSquarePants.pth"

And here is the log:
"D:\AI-Programs\pinokio\api\rvc.pinokio.git\app\logs\SpongeBob SquarePants"

[brittle wing]

so

[rare goblet]

Ayo? @brittle wing level 2 !!!

[brittle wing]

its 44khz

is this bad?

[slate terrace]

Does anyone have the "applio (mangio-rvc-fork)" google colab working?

[brittle wing]

valueError: 40000 SR doesnt' match target 48000 SR

why

[hasty sandal]

so i was leaning mainline

and i come across this instruction "1. Upload voice model.
Open RVC's folder, go to the assets folder and put your model's .PTH file inside the weights folder."

i dont know how to make the PTH file (voice model) in the first place and i cannot find a guid on it

like i have the audio but idk how to train it into pth and index

[hasty sandal]

actully im assuming its using tensorboard

[rare goblet]

Ayo? @hasty sandal level 1 !!!

[hasty sandal]

though i cant get that to work since when i put the .bat file in the rcv folder and tried to open it the console didnt reconize anything and wouldnt work

also its in this folder

[low tangle]

is anyone else having trouble connecting to a gpu on colab rn?

not sure if that even happens or if theres something else wrong

just a bad time of day maybe?

[low tangle]

well ok i connected but theres another problem that i keep having. ive tried resuming training several times now and it always starts over from the beginning. im using rvc disconnected

oh i think its because i got disconnected ughh nvm

[brittle wing]

is there a vocal remover app? since the website only allows me to seperate the song 1 time a day

[low tangle]

uvr5

[royal marsh]

Hello any help?

[light trench]

I hit a limit "ERR_NGROK_725 Network bandwidth exceeded" whats going on?

[ocean beacon]

is everything in #1159513888199540817 up to date? I'm reading through the documentation now to better educate myself and not come back with nonsense questions

[acoustic garnet]

the only things getting outdated is reading guides on outdated forks of rvc

[rare goblet]

Ayo? @acoustic garnet level 8 !!!

[rich brook]

how do i make this work on like roblox or discord

[rare goblet]

Ayo? @rich brook level 1 !!!

[ocean beacon]

Thank you

[rich brook]

hey you know how to make the emulator work in discord or roblox?

[hazy thunder]

uhh

whats with the big drop

[acoustic garnet]

mode collapse?

bit of a regular occurance id say

[ocean beacon]

So using weights.gg to create covers seems to automatically censor naughty words in the song. I'm assuming that directly using an RVC colab space would get around that?

[low tangle]

thats hilarious wtf

yes it would get around that

[dapper fable]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[ocean beacon]

Can't say the no no words I guess

[dawn nebula]

change dataset

[hazy thunder]

wait why

[rare goblet]

Ayo? @hazy thunder level 7 !!!

[hazy thunder]

my dataset is perfectly fine

[low tangle]

how long is it

[hazy thunder]

its about 11 minutes long

[dawn nebula]

dm me a short clip

[low tangle]

i ran into a padding size error when i was training with mangio on my own gpu because the files in my dataset were all about half a second long or shorter. would i likely have the same issue with rvc disconnected?

i tried putting them all together in succession but it created a very very choppy result during inference although it had a ton of promise.

my only idea that would get around the choppiness and the padding size issue was adding silence to the tracks, but theres just no quick way to do that when the dataset consists of over 20,000 of these very short clips, never mind the fact that doing that would add 5.5 hours to my dataset

i have another option, and im using it right now, but it's not ideal. so yeah the question is would rvc disconnected be able to use such short clips or is there just no way to make that work

i can also just try it later but id rather not use computing units if u dont have to

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[fluid topaz]

does applio has merge model feature? like the one in okada, because model created by applio can't be merged via okada, its missing "Info" metadata

[dawn nebula]

u dont have to slice

[low tangle]

ok ill try it, thank u

[brittle wing]

how do i make it so it can talk at a wider range of pitches

[low tangle]

id think you need a wider range of pitches in the dataset

[rare goblet]

Ayo? @low tangle level 5 !!!

[low tangle]

:-/ welp

[dawn nebula]

combine them

to a big one

[low tangle]

i did the first time but if theres no silence in between the result is very choppy so im trying to work out a different solution

[rare goblet]

Ayo? @brittle wing level 7 !!!

[fluid topaz]

sounds like crowd, try uvr crowd model

[rare goblet]

Ayo? @fluid topaz level 13 !!!

[fluid topaz]

[tribal sky]

has anyone tried both RVC-Project and Ilaria RVC?
i am considering trying ilaria because i'm curious how different it is from the og(?) rvc, i'm just not sure if i should keep my hopes up xD

[zinc anchor]

Ilaria RVC only does inference

[tribal sky]

ah i was referring to the beta that was uploaded in dev updates, i read that it also has training

[zinc anchor]

Oh nice, I haven't tried that one yet so I can't tell you much about it

[pastel oak]

@proven hill not sure if you mind pings - any major differences to the beta vs rvc-project when training?

[proven hill]

for the moment its offline and will be posted very shortly as 1.0

anyway the difference is a very much better ui, and slightly faster training times

[tribal sky]

ah i see
thank you

[rare goblet]

Ayo? @tribal sky level 1 !!!

[fluid topaz]

what batch size should i use when training with 4090?

[latent kettle]

I think it depends on your dataset length.

[molten trout]

How is this allowed but hitler isnt?

[fluid topaz]

the longer dataset the bigger batch size?

[latent kettle]

yes. for 2-3 hrs of dataset I suggest 16

[fluid topaz]

got it thanks

[proper shale]

You're just not allowed to ask around for it

If you wanna search for it that's your own problem

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[modern ravine]

thanks for the heads up. Downloading it before it gets deleted lol

[proper shale]

💀

dude

[pastel oak]

Mad

[grizzled forum]

how do i resume training on applio?

[odd shale]

🐢

[pastel oak]

🥬 🐢

[odd shale]

Lettuce and tutel.

[pastel oak]

Do you think tutel can eat that big of a lettuce I sent?

[odd shale]

Ofc it can.

[pastel oak]

Will it be hungry for more lettuce?

[odd shale]

My tootle always is hungry for lettuce.

As much as Ryanz' greek tutel.

[proper shale]

just insert the same settings as before, dont preprocess or feature extract again, keep training

[odd shale]

There's any applio colab?

[proper shale]

yeah

[odd shale]

Link plz.

[proper shale]

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

[foggy pebble]

Is there a free alternative to elevenlabs for text to speech instead of RVC?

[marsh sigil]

I have a question regarding AI voice training. Is it recommended to use ogg or wav files?

[proud elbow]

wav for compatibility and fidelity

[proper shale]

You're probably gonna like GPT-SoVITS

[proper shale]

WAV, because it's uncompressed + anything supports it

[marsh sigil]

That's acknowledgeable. Thanks for letting me know

[rare goblet]

Ayo? @marsh sigil level 1 !!!

[proper shale]

But if you have OGG only it's okay to use it, I guess

RVC preprocessing will convert it to wav regardless

But it may mess up

[proper shale]

https://docs.aihub.wtf/tts/gpt-sovits/

Hope you like it

[foggy pebble]

what about whisper AI?

[rare goblet]

Ayo? @foggy pebble level 2 !!!

[marsh sigil]

Thanks for letting me know as well 🙂
Hope you have a great time 😃 ✌️

[foggy pebble]

where's the mention of text to speech?

[proper shale]

GPT sovits is text to speech so...

[foggy pebble]

i didn't see text to speech screenshots and only labelling of samples

[proper shale]

whisper is like... transcription stuff? iirc

[proper shale]

well yeah that's what you'd do to train a model, but if you wanna use it you can use it just fine using the colab that was linked iirc

or this one, if it works:

https://colab.research.google.com/github/X-T-E-R/GPT-SoVITS-Inference/blob/stable/colab_webui.ipynb

[hot lagoon]

General question, but do y’all leave? The back vocals as original or something else I just want to see what is the most most common for back vocals for covers

[pastel oak]

I dont understand the question, whats the context? Like dataset or infering or

[hot lagoon]

Oh, I mean infering

[proper shale]

just keep as original

[hot lagoon]

K

[snow rose]

when i try to run easier guy i got this:
D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0>C:\Users\Adam\AppData\Local\Programs\Python\Python312\python.exe EasierGUI.py --pycmd C:\Users\Adam\AppData\Local\Programs\Python\Python312\python.exe --port 7897
D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0\EasierGUI.py:1789: SyntaxWarning: invalid escape sequence '\c'
value="E:\codes\py39\test-20230416b\todo-songs",
Traceback (most recent call last):
File "D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0\EasierGUI.py", line 1, in <module>
import subprocess, torch, os, traceback, sys, warnings, shutil, numpy as np
ModuleNotFoundError: No module named 'torch'

D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0>pause
Press any key to continue . . .

(im new to this kind of stuff)

i runned install easy dependencies

and install easy gui

before

@flint geyser

[rare goblet]

Ayo? @snow rose level 1 !!!

[flint geyser]

Yes?

[snow rose]

.

[flint geyser]

ah.

try pip install torch

torch doesn't seem to be installed in the venv

[snow rose]

hmm

[flint geyser]

py -m pip install torch

try that

[snow rose]

okay its downloading

i thought i have torch already

now i get this

D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0>C:\Users\Adam\AppData\Local\Programs\Python\Python312\python.exe EasierGUI.py --pycmd C:\Users\Adam\AppData\Local\Programs\Python\Python312\python.exe --port 7897
D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0\EasierGUI.py:1789: SyntaxWarning: invalid escape sequence '\c'
value="E:\codes\py39\test-20230416b\todo-songs",
Traceback (most recent call last):
File "D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0\EasierGUI.py", line 2, in <module>
from mega import Mega
File "C:\Users\Adam\AppData\Local\Programs\Python\Python312\Lib\site-packages\mega_init_.py", line 1, in <module>
from .mega import Mega # noqa
^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Adam\AppData\Local\Programs\Python\Python312\Lib\site-packages\mega\mega.py", line 18, in <module>
from tenacity import retry, wait_exponential, retry_if_exception_type
File "C:\Users\Adam\AppData\Local\Programs\Python\Python312\Lib\site-packages\tenacity_init_.py", line 451, in <module>
from tenacity._asyncio import AsyncRetrying
File "C:\Users\Adam\AppData\Local\Programs\Python\Python312\Lib\site-packages\tenacity_asyncio.py", line 33, in <module>
class AsyncRetrying(BaseRetrying):
File "C:\Users\Adam\AppData\Local\Programs\Python\Python312\Lib\site-packages\tenacity_asyncio.py", line 41, in AsyncRetrying
@asyncio.coroutine
^^^^^^^^^^^^^^^^^
AttributeError: module 'asyncio' has no attribute 'coroutine'. Did you mean: 'coroutines'?

D:\Gry\A_other\Vocal spliter\Mangio-RVC-v23.7.0>pause
Press any key to continue . . .

[snow rose]

@flint geyser

[flint geyser]

...

read the docs..

[snow rose]

i readed it...

[flint geyser]

..

try reinstalling mangio

or use something else

[snow rose]

WHYYYYYYYYYYYYYYYY

@flint geyser i reinstalled it what now

[flint geyser]

use ilaria rvc

[pastel oak]

Did you download the code or the binary release btw

[flint geyser]

oh yeah

that

[snow rose]

https://github.com/Mangio621/Mangio-RVC-Fork/releases/tag/v23.7.0

mangio rvc infer bat

then i runned it and it gived me zip file

[pastel oak]

How big was the zip file in GB

[snow rose]

2,37Gb

[pastel oak]

Anyway, if it causes issues, download mainline instead. You got a Nvidia gpu or amd?

[snow rose]

nvidia

[pastel oak]

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006Nvidia.7z

[snow rose]

@pastel oak i downloaded it

[pastel oak]

Ok

Run go-web

[snow rose]

gimme sec beacuse i installed it on HDD

[rare goblet]

Ayo? @snow rose level 2 !!!

[snow rose]

and how to use it

thats why i wanted easier GUI

[brittle wing]

someone knows why when i try to log in google chrome the website 127.0.0.1 doesn"t work?

[pastel oak]

https://docs.aihub.wtf/rvc/local/mainline/

[rare goblet]

Ayo? @visual flint level 4 !!!

[visual flint]

Is the offline version of Applio GUI the same as the browser/colab version @tepid bison ? Need the pitching and TTS options

Also i dont suppose its possible to save the cloud one locally to drive at all? i know u said the easyGUI one yeets ur storage, i assume applio is the same?

[half cove]

i assume you meant to ping me?

[snow rose]

@pastel oak isnt that model worster than the other one?

[half cove]

local should be the same

and it saves in your folders so no cloud storage needed

[pastel oak]

what model

[snow rose]

.

this

[pastel oak]

That is an application not a model

It is not worse

[visual flint]

sweet thnx. i dont suppose its possible to save the prerequirements on the drive one?

[half cove]

hm?

[visual flint]

what i asked u about a while ago. i suppose saving the install process would still yeet ur storage?

[half cove]

ahh

i dont think any of the colabs rn have it

also prob slower cuz gotta copy it all

[obtuse cape]

Yo I know this is off-topic, but I need your help so I can get my vocals to my rvc. Since it's self-made there's no YouTube video for my vocals, but I can load it through Google Drive.
It says I have to copy the file path from my drive, is there a specific way to do it since there's no official way for that?

The RVC I'm using is a CoverGen No UI

nvm I found it
since I linked my drive to the RVC I could just browse the file and copy it's path there

[dark harness]

when i pressed the run-applio.bat it said this:
Please run 'run-install.bat' first to set up the environment.

what to do

[half cove]

where did you download from

[dark harness]

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

[rare goblet]

Ayo? @dark harness level 1 !!!

[half cove]

wrong ping

[half cove]

prezip or from github

[dark harness]

zip

[snow rose]

@pastel oak im just asking out of curiosity do you recommend changing any settings? most people say default are the best but from what i can tell you know i think more about this application

[half cove]

if its the full zip that warning should not come up

[obtuse cape]

don't worry about it

[half cove]

send where you downloaded from

[dark harness]

https://huggingface.co/IAHispano/Applio/blob/main/Compiled/ApplioV3.0.8.zip

[half cove]

weird that should have it all

it is an old version

try a newer version or an exe version
https://huggingface.co/IAHispano/Applio/tree/main/Compiled

[dark harness]

3.1.1 with pickle?

oh wait thats zip

[pastel oak]

Default ones are usually good enough unless you need something specific. The docs guide briefly explains what each thing does

https://docs.aihub.wtf/rvc/resources/inference-settings/

[snow rose]

thank you

@pastel oak there is no Transpose setting that is in this document that you sent where you can control pitch

[pastel oak]

Well I have the transpose setting so im not sure what youre looking at

[snow rose]

can i sent you the image on dm?

[pastel oak]

Send in here

[snow rose]

wait its tottaly in diffrent section but i finded it

[glad ether]

Hey so i'm using the common google collab or better said i'm trying to get it work buuuuuuut i get this error:
[Can't post pictures somehow?]
`JSONDecodeError Traceback (most recent call last)

<ipython-input-15-77d3f5a86bb5> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:`

/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx) 353 obj, end = self.scan_once(s, idx) 354 except StopIteration as err: --> 355 raise JSONDecodeError("Expecting value", s, err.value) from None 356 return obj, end

my hope are not high that someone can help me but maybe?

its on the part trying to download the model | step 2

i was trying to download this alastor model huggingface.co/ScruffyRVC/RFCRVCV2/resolve/main/alastorfinal3.zip

[half cove]

this is an old colab

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[half cove]

look above for a newer one

[glad ether]

oh thank youuuuuuuuuuuuu which do you recomend?

[rare goblet]

Ayo? @glad ether level 1 !!!

[glad ether]

damn i feel dumb now xD

[half cove]

try applio ig if you want to make covers

we have a guide on our docs

[glad ether]

ty <3

[half cove]

https://docs.aihub.wtf/rvc/cloud/applio-colab/

[glad ether]

nice! thank you so much <3

[wispy burrow]

@half cove Hey.
I was suggested to ask you about an issue/error I'm having with with Ilaria RVC Beta.

[half cove]

whats the issue

[wispy burrow]

The very moment I start training I get this error.

Traceback (most recent call last):
File "C:\Users\User\Documents\AI\Ilaria_RVC_Beta_2\infer\modules\train\train.py", line 14, in <module>
hps = utils.get_hparams()
File "C:\Users\User\Documents\AI\Ilaria_RVC_Beta_2\infer\lib\train\utils.py", line 368, in get_hparams
config = json.load(f)
File "json_init_.py", line 293, in load
File "json_init_.py", line 346, in loads
File "json\decoder.py", line 337, in decode
File "json\decoder.py", line 355, in raw_decode
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

[half cove]

this happens with all pretrains?

[wispy burrow]

Oh, I haven't tried them all.
I only tried with RIN so far.

[half cove]

ilaria seems to have deleted the prezip

its prob the uhh

its prob a typo lmfao

try another pretrain ig

[brittle wing]

Is that overtrained

[wispy burrow]

Tried them all now.
I get the same error for all of them.

[half cove]

shi

[pastel oak]

It looks like youre nearing another low point, so would keep training

[half cove]

give me your
configs/config.py
and infer-web.py

[brittle wing]

Do you see someone's face in the photo or I cropped it out successfully

[pastel oak]

No face

[brittle wing]

Ah okay nice cause I was watching a video of the person I'm making a model of

[keen notch]

hello. what should i do with dataset_folder ?

[tough fiber]

how can i sell my voice models

[fluid topaz]

put it on website? anyway you can't post here because you're not a model master

[tough fiber]

already i have website im just asking how can i get model master thing

[fluid topaz]

well idk about that

[hot lagoon]

how do I use voice synthesizer for my covers

[low shard]

first of all you need to be a model maker

https://docs.aihub.wtf/extra/model-maker-role/

after u got model maker, u gotta make some very high quality models and ofc it will take time for this all, you can use /jointeam and apply for model master

[brittle wing]

im trying to make a song cover with Chester's voice, but it sounds terrible, what i should do to enhance my music?

[fervent sun]

this is my first time attempting to make an AI mdoel, on the RVC v2 disconnect what do I do once I have made it to the training section?

I don't see where I would upload files for it to train the voice model

[pastel oak]

https://docs.aihub.wtf/rvc/cloud/rvc-disconnected/#training-

theres a load dataset section where you upload your dataset as a zip file

[fervent sun]

ohhhh okay see i just made an empty zipfile lmaooo i didn't realize that's where you stored the files for it to train. thank you!

[pastel oak]

bro was trying to train some air 🧢

goodluck with da training

[fervent sun]

🤣

thank you!

[pastel oak]

Define "sounds terrible"

The quality depends on:

• the voice model
• the acapella you infered with

If its a woman singing, you'd obviously have to go down in transpose to fit chesters vocals.
Else making sure that the acapella is as clean as possible

[brittle wing]

it's the acapella of disturbed - the animal song

[dapper forge]

Hello, does anyone know how to make the AI ​​also sing the chorus and not leave them with the original voice? Will there be any configuration for that?

[pastel oak]

That does not make sense. What application are you using to make the covers

[dapper forge]

I use ai cover gen what happens is that the song sings everything well but there are some parts of the song where the voice does not sound and the original voice sounds I am using translator (":

[pastel oak]

Ah I unfortunately dont know anything about cover gen, if someone else knows theyll reply to you

[dapper forge]

Thanks(":

[rare goblet]

Ayo? @dapper forge level 1 !!!

[queen compass]

Any opinions on Mangio RVC vs Applio RVC vs Original Github RVC? Like what's the gold standard

in regards to creating ai models

[viscid moss]

If you are talking about the "quality" of the models, it is the same for all so... nvm

[queen compass]

I see, that makes sense, Which one would you say has the most easy to understand user interface?

[viscid moss]

Mangio has a fairly simple interface, the Mainline (Original Github RVC) is a little more complex because it has extra functions, but it is still quite simple.

https://docs.aihub.wtf/rvc/local/mainline/

U can check our docs for more info, so u can decide

[queen compass]

I see, thanks for the info, i'm going to go with mangio in that case since i've worked with programs that are quite similar to it

[proud elbow]

try separate using UVR

[fervent sun]

are studio session vocals a good way to train an rvc model?

[rare goblet]

Ayo? @fervent sun level 2 !!!

[brittle wing]

how do i up the pitch for real time rvc?

the model sounds raspy

from MTF

plz

[rare goblet]

Ayo? @brittle wing level 1 !!!

[queen compass]

I just had the same problem

[low tangle]

does anyone know what to do about this

ive never had this issue

[fervent sun]

@proud elbow how many epochs is it supposed to go through?

@pastel oak ^ ?

[low tangle]

chatgpt says theres an issue with the torch import?

[gusty kestrel]

@brittle wing, I have found 10 results that match your search!

rainbowzigzag2006

GIRL FEMALE

• Uploaded: <t:1702937650:d>
• Server: AI Hub
• Likes: 0

Epochs

280

Technology

RVC

Algorithm

Rmvpe

Requested by x1aa · Apr 19, 2024, 2:08 AM

[viscid moss]

yep

[fervent sun]

thank you

[viscid moss]

Change the tune

[fervent sun]

i also figured out i didn't have tensor on & that measures the right amount of epochs

so that answers that question

1 other question i did have is how to properly upload the model onto huggingface because i couldn't get it to work right

[viscid moss]

What colab are you using?
Maybe you used up your daily GPU usage in Colab

[viscid moss]

https://docs.aihub.wtf/essentials/voice-models/#uploading-to-hugging-face
Check this

Our guides website

[fervent sun]

i didn't know what to put the license as so i put unknown that might have been why

thank you! @viscid moss

[viscid moss]

doesn't matter to much, but is better to select openrail

[fervent sun]

i probably downloaded it wrong tbh still trying to learn

[low tangle]

rvc disconnected, i just logged on

[viscid moss]

Try this:
Go to runtime > disconnect and delete runtime

And then start again

[brittle wing]

I'm running on a Mac with an M2 and trying to use so-vits-svc-fork. I used one of the voice models, but I ran into this: "utf-8' codec can't decode byte 0x80 in position 64: invalid start byte." I usually use .json instead of .index. If someone can help with the issue, it'll be great;y appreciated

[low tangle]

didnt get the chance to try bc its working now but thank u ill try if it happens again

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[rapid spade]

how do you update the ai thing since it is giving me error

[brittle wing]

Yeah same, I downloaded current version and getting errors while training.

Folder paths of the datasets are fine but still getting weird python errors.

[low tangle]

what do u mean by "ai thing"

[low shard]

Wdym with ai thing? And are you following our docs ?

[low shard]

You following the ones from our docs?

Also pls be specific abt errors

[low shard]

Uhh maybe kaggle

Ye ofc everything online has gpu limutations for free users, gpus are expensive so they can't js give it free no limitations for everyone

It has like 48 hours of free gpu weekly

Honestly never used kaggle but surely as its basically 2 days of gpus

What matters the most is that the dataset is very well cleaned, so if the 1 hour one is like very good cleaned you could use that, just jnow there shouldn't be that much huge difference but its good

Btw for kaggle u could use https://www.kaggle.com/code/hinabl/multinotebook-ai-public

As in the site is referred as vocal model, yea

Yw

It should be still updated i think

Oh btw for kaggle u need to put a phone number to use their gpu

Its a requirement for not making their gpu abused like it happens to colab with alt acc lol

[sacred nimbus]

is there aa male version?

[low shard]

Wdym ?

[sacred nimbus]

like a rappers voice

[low shard]

There are for sure rvc rappers voice models

you can search rvc ai voice model:

• #1175430844685484042
• #🔍│find-models
• https://weights.gg/
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models

[worn zinc]

Wanted to stop training a model and continue it later on colab (rvcDisconnected), how exactly do I do that?

(Ping me in a reply for this, appreciate it)

[rare goblet]

Ayo? @worn zinc level 5 !!!

[worn zinc]

(ayo that got me back to verified again noice)

[pastel oak]

Make sure to save the G and D paths in logs

[worn zinc]

2333333?

[pastel oak]

ye

[worn zinc]

alright, got it

[pastel oak]

then when you resume training you put in the same data as usual, process the dataset, feature extraction, then choose those g and d paths on the pretrain slots

[worn zinc]

If I saved the preproc data in Google Drive, that's the only thing I need to load (besides the metadata) right

Like the training vars (name, model arch, pitch algo) pretrain and dataset?

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[torn depot]

how do i make a voice model sound more energised?

[burnt timber]

hey looking for someone who can do a high quality french voice of a girl, will pay

[low shard]

#1159289738314919936 or #1191429836321849435

[silver notch]

im using an rvc voice changer but its not w-okada and it keeps saying pth文件路径不可包含中文

nvm i fixxed it

[brittle wing]

can i get website models ?

[proper shale]

?

[brittle wing]

where from i get models

[low shard]

you can search rvc ai voice model:

• #1175430844685484042
• #🔍│find-models
• https://weights.gg/
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models

if there isnt one, #1159289738314919936 or make it yourself with our docs guides https://docs.aihub.wtf

[proper shale]

^

[viscid moss]

https://colab.research.google.com/github/Eddycrack864/UVR5-NO-UI/blob/main/UVR5_NO_UI.ipynb

My new colab :3

[fervent crystal]

How long does generating an AI song file in RVC GUI take? Sorry if this is a stupid question, it's my first time

what if my processing mode is CPU

[rare goblet]

Ayo? @fervent crystal level 1 !!!

[pastel oak]

Just use ilarias rvc at this point

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[pastel oak]

the Ilaria Google Colab

If you want a step by step tutorial in case youre not familiar:

https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/

[fervent crystal]

alrighty thanks

I'll try it out

[heady valve]

How do i text to speech on mobile using rvc

[proper shale]

you can just do tts on ElevenLabs and then convert the result in RVC

https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/ on Ilaria RVC, so you can do the RVC part of the thing

[heady valve]

🫡thanks

[fervent crystal]

ok I figured it out and it worked

thanks for the help

[proper shale]

yw :)

[wintry grotto]

where did the downloads go for ilaria?
https://huggingface.co/TheStinger/IlariaRVCBeta/tree/main

[rapid spade]

Nvm the software was doing one of its moments it works now

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[low shard]

i think the ilaria mainline download got moved to github as you see here #📰│dev-updates message

@proven hill u may want to delete that hf repo model btw as its empty

[proven hill]

it will be re-updated

[low shard]

Oh okay

[wintry grotto]

its only got the source code there

[low shard]

no, read the instructions, theres a download link

[wintry grotto]

when i try to download that it says entry not found

[low shard]

@proven hill btw u may wanna change that download link in there

[half cove]

seems like she deleted the prezip

ig for a new one

[wintry grotto]

[proven hill]

a 1.0 will be out, just wait

theres literally no rush

[low shard]

i think going on the github repo, clicking code and download as zip would work for now

[proven hill]

you to venv it

[kind gyro]

Hey anyone wanna join with me on my song project?

Hello

[rare goblet]

Ayo? @kind gyro level 1 !!!

[kind gyro]

@rare goblet Yup just joined didcord

For the first time

Arcane wanna make a song

I want a good singer

[brittle wing]

Is that overtraining

Nope the real question is the model is about to finish it's training and I should quit anytime soon?

[low shard]

its a bot lol

[pastel oak]

Your most recent OT is at 35k or 37k I cant quite tell which one of those, and youre at 41k training atm. I would keep training just in case it reaches another one. If nothing comes in the next 5-10k steps you could assume it has finished at 35/37k if you dont want to keep training

[brittle wing]

It's not overtraining it created another point

[pastel oak]

Alright

[proper shale]

No

[proper shale]

I mean if that's what you want, sure

[brittle wing]

No I won't I truly want it to turn out even better

[proper shale]

Based

[proper shale]

So you have a long long road to go I think

Have you tested the model yet?

Also, maybe give us mel and kl ;)

[brittle wing]

Tbh I trained the same dataset on RVMPE it took only 380 epochs to complete yet mangiocrepe takes more I think

Cause it picks up more details

[proper shale]

Don't think it's related to that

Maybe the rvmpe version wasn't so completw

Also, mangio is more noise sensitive... so you do need a good denoised dataset

[brittle wing]

An admin said crepe picks up more detail.

[rare goblet]

Ayo? @brittle wing level 12 !!!

[brittle wing]

I trained my first model like this

[proper shale]

and more noise

and tbh not worth it 99% of the time

[snow rose]

false

[brittle wing]

Nah

[brittle wing]

It's worth it RVMPE is monotone

[proper shale]

yes, it does

unless u use high crepe hop

[brittle wing]

Well I used default 64

[proper shale]

i will confess that, if you have the right dataset, crepe might benefit from it, as @golden creek pointed out once in their testing

however, rvmpe is just better for most stuff, denoised or not

that's why everybody recommends rvmpe, because it just works fine for everything

[snow rose]

and what do you think about harvest

[brittle wing]

No it's not.

[brittle wing]

This one's bad I think

[proper shale]

it was good for its time

sadly, being cpu bound... not worth it nowadays

its rvmpe v1 if you think abt it

[proper shale]

if you have a specific use case it fails on, tell me

because i have genuinely not seen one in quite a while

[snow rose]

Sukuna singing specialZ

[proper shale]

ah yes, the classic

[brittle wing]

Litsa the dancer told me crepe picks more detail and also mangio crepe is better

Woof

[proper shale]

look, knowing how that song sounds

it's

but yeah i have seen once a kpop song that only worked well enough w harvest because it had a good low range

[snow rose]

somehow idol sounds good with most models so i thought specialZ would not be a problem

[proper shale]

Litsa is correct mostly

If you have a low crepe hop, yes it'll pick up more detail

But also more noise lol

And if you have a noisy dataset, it screws up

[brittle wing]

Jhope changes his voice frequently different intonation and flows that's why I'm using it

[proper shale]

RVMPE is best for most stuff because of this

[brittle wing]

You think I haven't denoised my dataset?

[proper shale]

It doesn't need the best dataset to work well

[proper shale]

If you did, that's great

But even so, it's not really worth it to use crepe over rvmpe

[snow rose]

Any tips to make it sound better? Beacuse you know the song i can tell

[rare goblet]

Ayo? @snow rose level 3 !!!

[proper shale]

Unless you reeeeally want that extra quality

[brittle wing]

Nah I have tried to convert the same song with the crepe or RVMPE one the crepe one still beat it to it

[proper shale]

I'd say try different pitch extractors, dereverb if you haven't yet... and mess with the pitch options too

[proper shale]

That's inferencing and not training, in inferencing anything can happen

[brittle wing]

Inferencing is for proof

[brittle wing]

How it sounds

[snow rose]

i tried most things

[proper shale]

Literally had harvest beat both RVMPE and crepe once

[proper shale]

Does that make harvest necessarily better?

No.

[proper shale]

did u use karaoke separation too

[snow rose]

yup

[devout berry]

anyone facing an issue with RVC not saving settings ?

[proper shale]

If you wanna use crepe for inferencing, that's okay.

But for training, you'd probably never hear the difference unless it's an extremely specific use case

[proper shale]

Hm?

Explain this

[brittle wing]

Depends

Maybe

[snow rose]

its even worser with it

[brittle wing]

On the voice

[proper shale]

oh damn

yeah it might be the song

[brittle wing]

I can't ask an admin

[proper shale]

No I was asking @devout berry

Dw

[proper shale]

Because it varies, yeah

[devout berry]

pressing save settings wont save any of the settings like tune index for models

[rare goblet]

Ayo? @devout berry level 1 !!!

[proper shale]

In extreme cases some can be better than others, HOWEVER, rvmpe is generally the best

[proper shale]

Oh damn

It should do that... haven't tested saving settings in a long while

[devout berry]

yah so everytime i swap between models the tune and index get reset

[proper shale]

I think the folders inside model_dir (aka your models) have the settings for each model

Wait let me take a look

[devout berry]

nothing there even after pressing save

[proper shale]

Every model you upload there should have a params.json file with the pth and index stuff

[devout berry]

nope nothing forsome reason

[snow rose]

@proper shale is trying De echo worth installing it?

[proper shale]

onnx file too?

[proper shale]

ehhhh idk tbf

[devout berry]

i extracted onnx

using the export to onnx button

but there is no json file for settings or something like that

[proper shale]

try save setting and see if it makes a json

else, I'd just make one myself

[proper shale]

wait... you don't need export to onnx

if you're using GeForce experience

unless you have an older nvidia gpu

[devout berry]

was just testing it didnt see any difference so just went back to using the original model

i have rtx 2060 idk where does that put me in the performance chart for rvc

[proper shale]

you're using the wrong version for your gpu too so maybe that caused smth

[proper shale]

yeah that can run rvc just fine

[pastel oak]

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

[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 💖

[pastel oak]

heres the download link for the right wokada version

[proper shale]

^^

[devout berry]

isnt it the same one u sent ?

oh im using directml

[pastel oak]

After I had one guy who failed to download the right version after telling him via the ai help thing I just gave up and always send the link directly ever since 😂

[proper shale]

yeah directml version is for amd

[proper shale]

what

HOW

[devout berry]

ooo well blame that youtube video i watched

[pastel oak]

AI Hub youtube tutorials coming soon hopefully

[devout berry]

saying i need this version no matter what gpu i have

[pastel oak]

A particular trust in people broke that day

[devout berry]

but what difference does it make using directml or the nivida one ? better quilty and faster response ?

[rare goblet]

Ayo? @devout berry level 2 !!!

[proper shale]

add no extra spaces here, also delete parenthesis

{
"slotIndex":(insertnumberhere),
"voiceChangerType": "RVC",
"name": "(insertnamehere)",
"description": "",
"credit": "",
"termsOfUseUrl": "",
"iconFile": "(insertifyouwant,elsedelete)",
"speakers": {
"0": "target"
},
"modelFile": "(insertfilenamehere)",
"indexFile": "(insertindexfilenamehere)",
"defaultTune": (inserttunenumber),
"defaultIndexRatio": (insertratiohere),
"defaultProtect": 0.5,
"isONNX": false,
"modelType": "pyTorchRVCv2",
"samplingRate": (insertsamplerate),
"f0": true,
"embChannels": 768,
"embOutputLayer": 12,
"useFinalProj": false,
"deprecated": false,
"embedder": "hubert_base",
"sampleId": "",
"version": "v2"
}

im dumb

yep no

[devout berry]

what should i name the json file tho ?

[proper shale]

nvidia one works better for nvidia GPUs

[proper shale]

params.json

[devout berry]

i mean yah ofc but its weird that even the amd one works fine 👀

i will try the nivida one now

[proper shale]

DirectML works fine for nvidia too so

[snow rose]

@proper shale im reading docs and it says that .wav is better than mp3 do you think it makes significant changes to split it from start?

[proper shale]

unless you have a .wav source of SpecialZ, no

[snow rose]

i can download from youtube in .wav format

[proper shale]

yeah that is not technically a wav source

since yt stores on opus (which compresses stuff)

[devout berry]

oh yah i can confirm the nivida version sounds better indeed

and it saves now

https://tenor.com/view/lets-go-gif-drake-gif-26307972

[proper shale]

Woo

Glad to have helped 🫡

[snow rose]

@proper shale im trying to do specialZ for like 2 hour and i think its impossible

[icy lance]

Hey guys, The "HF Space" mentioned in this document (https://docs.aihub.wtf/rvc/resources/datasets/) keeps giving me an error when trying to insert my data. It's only around 1 minute of data as to not enter too much as listed above.

I'm doing this to get the sample rate (Although I don't exactly know what that is, but it says I need it for a later step). Anyone got any ideas why it isnt working? Thanks in advance!

[low shard]

@proven hill there's an error with https://huggingface.co/spaces/TheStinger/Ilaria_Audio_Analyzer js tried myself too

[proven hill]

oh

its probably because we tested few things...

[low shard]

idk i just tried with random audios and all the audios give error

btw i see a pull request

u may wanna check that ig to see if it fixes it

[icy lance]

Is there any tutorial on how a .pth file is created from your dataset? In this document (https://docs.aihub.wtf/essentials/how-to-make-voice-models/) Step 1 teaches you to create a .flac file from audacity. Step 2 in the local tutorial it assumes you already have a .pth file.

I have no idea how I would generate this and there doesn't seem to be any tutorial on it.

[low shard]

on the step one, theres a learn how hyperlink where it redirects u to how to make your dataset, for the step 2, its js telling u all the rvcs u can try, just pick any of those ( be sure that it has training) and check the training part instead of inference

what pc gpu u got?

[icy lance]

I am really bad with locating my specs, could you tell me where I can find this?

[ornate wasp]

Anyone ever use RVC with Vast.Ai? I got stable diffusion running on it and was wondering if anyone has a good docker template/image for it

[ornate wasp]

Easiest place to find that would be the task manager under the performance tab

[icy lance]

Thank you! Seems I have the "NVIDIA GeForce GTX 1070"

[rare goblet]

Ayo? @icy lance level 1 !!!

[low shard]

do ctrl + shift + esc and task manager should appear, show a screenshot of the gpu on performance tab

[low shard]

whats vast.ai?

is it like google colab? like whats their free limits ?

[icy lance]

There you go, hope this helps

[low shard]

yea that gpu is not ideal for training, check this https://docs.aihub.wtf/rvc/cloud/rvc-disconnected/ or any other google colab inside the cloud folder inside the rvc folder. google colab is a cloud service that lets u run code on google pc gpu

[ornate wasp]

Cloud GPU service. It's like collab and paperspace and all those. No free tier, but it's the cheapest I've found. TensorDock is also cheap but it's a pain in the ass to set up. Vast works great tho, they have tons of premade templates for Stable diffusion and other programs.

Yesterday I paid 15 cents an hour for an a4000

[icy lance]

Thank you! I'll check it out

[ornate wasp]

They don't have any templates for RVC so I figured this would be the place to ask

[low shard]

is there a free option or just cheap?

[ornate wasp]

Just cheap, like stupidly cheap. So cheap that I was afraid to even try it. But I'm glad I did, works so good. Setting it up was as easy as collab

[ornate wasp]

As for TensorDock that also has no free tier, but if you ever want to try it out let me know and I'll let you use the rest of my credits on there. I was unable to get RVC running on it and don't want to credits to go to waste

[low shard]

Looks very good even if prolly most people wont use it bc even if its cheap asf its not free and most people here are minors 😭

[low shard]

wait u sure? how many credits u even got lol

[ornate wasp]

Like $1.20 left on it, I only loaded $5 to test it out so it wasn't much of a loss

[low shard]

ohh dw

[ornate wasp]

But yeah Vast.ai is much better than TensorDock simply because of the Docker templates. 1 click and I had Stable diffusion running. Hopefully I can find an RVC Docker template/image so I can try that out

[red kayak]

u can just make a Jupiter notebook for applio

they have stuff precompiled in just one line if u git clone their repo

[ornate wasp]

Damn I was not aware of that, let me give that a try deal quick. Thank you!

[icy lance]

Hey guys! I'm using the RVC v2 Disconnected google collab and I think I am training my model! (Console showing epochs being trained)

A little problem, I don't know what to do once it's finished/ end the proces if its too far. I didn't really understand what to do from the tutorial. What I really want to know is.

1. How would I stop the training?
2. How do I know how to stop the training?
3. How do I export the results?

This is probably explained in the docs, but I just don't know for sure how to do it!

[proper shale]

1. just click on the stop button that was originally the play button, just make sure it's not saving an epoch

2. after 1 or 2 hours you should stop, and then return later

3. run the export model step

[icy lance]

Thanks for the response! I saw some code blocks talking about returning after training, so I guess I use those when I return later.

After stopping it for the first time, can I already export it to test the results?

[proper shale]

yeah, just run the export model step (i think the Huggingface one is even better so it speeds up the process) and use a colab like Ilaria RVC to test it out

[icy lance]

Thank you for the information! I'll see if it works!

[proper shale]

Right, let me know how it goes :)

[brittle wing]

how can i use voice ai? where is link

[proper shale]

We don't give support to voice.ai specifically, but we can help you with realtime voice changing using AI

https://rentry.co/VoiceChangerGuide should help

[brittle wing]

its just realtime?

[proper shale]

This one yeah, but if you want anything else (for AI covers, for example); https://docs.aihub.wtf/

[brittle wing]

thanks. is realtime voice change easy?

[rare goblet]

Ayo? @brittle wing level 1 !!!

[proper shale]

yeah, you just need a good GPU tbh

[brittle wing]

thanks again.

[proper shale]

glad to help :)

[ornate wasp]

Jupyter keeps asking for this Y/N bullshit but there's no way to input the y

Haha I got it!

Just had to add --yes and -y to a dozen lines

[gentle verge]

why is my models lagging and has a huge delay

im new so idk

[proper shale]

are you on AMD?

or what GPU do you have

[gentle verge]

yeah

[proper shale]

oh yeah you need to convert downloaded models to onnx

https://rentry.co/VoiceChangerGuide#uploading-models-amdintel-gpus-and-fixing-its-laggy-issue

[gentle verge]

oh

[proper shale]

and make sure you're using the right version of W-Okada for your GPU too, it's the DirectML version ;)

[gentle verge]

thanks

[proper shale]

yw :)

[dry prism]

I'm using the rvc web ui for training a voice changer, and attempting to get a cli working for it -- I can't use gradio api calls because I'm trying to run on condor, which does not give me permissions for networking, even whithin the docker container. It seems like setting the sample rate for preprocessing to 16000 makes training not work -- since there isn't really a 16k model or config to start with (or something along those lines). As a quick fix, I'm up-converting all my audio files to 48khz before running the preprocessing with 48000 as the setting. Is this wise? Would the training possibly get messed up by this upsampling?

[rare goblet]

Ayo? @dry prism level 1 !!!

[proper shale]

wouldn't really recommend upsampling tbh

do you only have 16k audio?

[dry prism]

only 16k, using part of the librispeech dataset

open-source dataset of audiobooks

[proper shale]

hm... can't you train on 32k like on webUI? 🤔

[dry prism]

would that work fine with 16k audio?

[proper shale]

yeah probably would work best

[dry prism]

when I run preprocess, would I have to run it with 32000 as the sample rate?

I really can't get it working with 16000 as the sample rate, something gets messed up

[proper shale]

yeah

[rare goblet]

Ayo? @proper shale level 96 !!!

[icy lance]

Wanted to let you know that I succeeded! 180 epochs trained on 11 minutes of me talking and it sound really great. Might train it a bit higher tomorrow with the continue feature. Do you recommend a certain amount of epoch or is it just trail and error? Most models I saw have around 300 epochs.

Thanks for all the help btw, and to the other people who helped me as well!

[rare goblet]

Ayo? @icy lance level 2 !!!

[proper shale]

Glad to know you've gotten good results!

There isn't exactly a recommended epoch count, because all models are different. I'd say keep training until you notice overtraining, or set a goal of 500/1000 epochs (of course, if not overtraining).

[icy lance]

Thanks for all the help, I'll try that later!

[fervent sun]

Would 6:00 of vocals be too much, like should I only feed the AI a small amount to train with or is there never too much?

[odd shale]

If your dataset is clean enough, 6 minutes can work nice.

It's matter of testing unless you're able to gather more dataset.

But if you only got 6 mins, go with that.

[fervent sun]

would you recommend more than 6 mins?

it's straight studio vocals

no effects or background noise

[odd shale]

Yeah, but if you are able to get more studio vocals, go and gather more.

I repeat, unless you're unable to get more.

[odd shale]

Make sure to remove reverb.

[fervent sun]

he only has 1 studio session available on youtube

no reverb either

i'll send the link & see what you think

https://youtu.be/aiyMEvJc_xg?si=J8pQ0wMk4yy9ksuT

[odd shale]

Umm.. I would recommend you to apply denoising.

[fervent sun]

Okay, because of the faint beat I'm assuming?

[rare goblet]

Ayo? @fervent sun level 3 !!!

[odd shale]

Yeah.

[fervent sun]

I just realized you could hear thay

thay

that**

Any software you can recommend I usually would use my phone to denoise

[odd shale]

yeah, things like static and faint beat can end causing artifacting when inferencing.

[odd shale]

Use denoise model in UVR.

[fervent sun]

Okay

Thank you

[odd shale]

Remember to play around with agressiveness settings on UVR when using denoise model.

Maybe you'll must test with different values till you don't hear that faint beat anymore.

[pastel oak]

🥬

[odd shale]

Hello Shad, my favorite W-Okader

[pastel oak]

Hello Leo, my favorite tutel helper

[odd shale]

How cute.

🐢 🇻🇪

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[quartz heart]

help

how do i tranfer google drive link to hugging face link ?'

and how do i use it

[gentle verge]

why wont my voice changer work

[rare goblet]

Ayo? @gentle verge level 1 !!!

[gentle verge]

i have it on

[viscid moss]

Download the model and upload to HF

https://docs.aihub.wtf/essentials/voice-models/#uploading-to-hugging-face

[viscid moss]

pls go to #🔍│help-w-okada

[ashen notch]

what is that?

what I can't use voice changer?

it's infinite trying

[viscid moss]

I'd suggest deleting the pretrain and model_dir folders inside MMVCServerSIO and restarting

[quartz heart]

got it thx

but how do i use it ?

[rotund cairn]

i really need your help its because i want to use a voice changer on tiktok live but samsung or iphone doesent allow it, i have a voice changer on my pc is there a way to connect my pc.s mic to my phone so all the audio that comes out of my pc mic comes out of the phone like a cable or just something

[viscid moss]

https://rentry.co/VoiceChangerGuide

[rotund cairn]

i really need your help its because i want to use a voice changer on tiktok live but samsung or iphone doesent allow it, i have a voice changer on my pc is there a way to connect my pc.s mic to my phone so all the audio that comes out of my pc mic comes out of the phone like a cable or just something
will send 7 dollars if someone can help

[quartz heart]

did u just rick rolled me for no reason?

[viscid moss]

nah ignore it

[quartz heart]

that voicechanger

i only want to play a song with a voice

[viscid moss]

mb

follow this guide instead:
https://docs.aihub.wtf/essentials/how-to-make-ai-cover/

[quartz heart]

i downloaded the voice model but its not showing it

😞

[viscid moss]

Did you press the refresh button?

[quartz heart]

yep

[viscid moss]

Show me the Download Voce Models tab

[quartz heart]

its good?

[viscid moss]

a

U need the .index file too

.pth and .index both files in a .zip

And u upload that .zip

[quartz heart]

so i need delete the pth for now ?

and then make both zip and uploade it?

[viscid moss]

yep

[viscid moss]

Just one .zip with both files inside

[quartz heart]

how do i delete the rest?

i uploaded now the zip

[viscid moss]

Just delete this file.

Click on it first

And then delete

[quartz heart]

tysmmm bro appreiciate it very muchhhhhhhh

its worked

🔥 🔥 🔥

[upbeat coral]

I keep getting this when running the illaria fork of rvc

echo Ensuring audio-separator is installed before startup. Ignore the following notices.
runtime\python.exe -m pip install --quiet audio-separator
runtime\python.exe -m pip install --quiet audio-separator[gpu]
echo Ilaria RVC Starting...

start /B runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897

[dense marsh]

can intel create covers locally?

[timid condor]

What kind of Virtual Audio Cable good and safe? please help

[dense marsh]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[quartz heart]

btw i get now an error like this you know how to fix it ?😅

[upbeat coral]

-local

[azure marsh]

🖥️ Local stuff

• 🍏 Applio, by IA Hispano GitHub
• Mangio-RVC-Fork, by Mangio621 Huggingface
• RVC Studio, by SayanoAI Huggingface
• AICoverGen, by SociallyIneptWeeb GitHub
• Replay, by Replay Team Website
• Original RVC, by the RVC-Project team GitHub
• GPT-SoVITS, by RVC-Boss GitHub

Need some help? 🤔

You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.

More commands: -audio, -colabs, -overtrain, -realtime, -rvc, -help

[viscid moss]

Is that colab? Can u show me the colab tab if so

[quartz heart]

u mean like this?

[viscid moss]

I mean this

[viscid moss]

hmmm

[quartz heart]

i thinking its well

💀

[viscid moss]

ya that's fine

lemme test rq

[quartz heart]

welp

ill try restart it again

[viscid moss]

ya, it's working on Hugging Face Spaces

[quartz heart]

can you do one song for me?

pls

i only need this

[viscid moss]

Ok, send me the audio file in DMs

But HF takes a lot of time to convert

[quartz heart]

OHHHHHHHHHH I GOT THIS BRO

NM

THX YOUUU

💞

😍

[quaint yew]

uhh

step 2 of the rvc isn't working

[proven hill]

you dont have python

you also dont have anything else installed

[fervent sun]

@viscid moss hey bro would you mind linking me to the page for explaining how to upload onto huggingface again

i can't seem to find it

and i can't figure out how to properly upload it 😭

[quaint yew]

what do i do in this case

"JSONDecodeError Traceback (most recent call last)
<ipython-input-5-2e074eafac12> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end

JSONDecodeError: Expecting value: line 1 column 1 (char 0)"

[pastel oak]

probably outdated colab

check for the newer ones

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[ashen notch]

thank you! I think this will resolve the issue

[pastel oak]

would suggest ilaria for inference, rvc disconnected for training

[fervent sun]

which one of these would be the correct file to zip for uploading onto HF?

[viscid moss]

https://docs.aihub.wtf/essentials/voice-models/#uploading-to-hugging-face

This, but u need a index file too

[fervent sun]

okay so i fucked that step up them

then*

[quaint yew]

ok thanksw

[rare goblet]

Ayo? @quaint yew level 1 !!!

[fervent sun]

the one right before training?

[viscid moss]

yep

[fervent sun]

i still have the colab open can i fix it still or is it too late

i just ran the index and it added this file

[upbeat coral]

I do, managed to fix that, now I am like 10 more issues deep

[rare goblet]

Ayo? @upbeat coral level 4 !!!

[upbeat coral]

@proven hill

2024-04-19 20:36:54 | INFO | configs.config | Found GPU NVIDIA GeForce RTX 3070
is_half:True, device:cuda:0
2024-04-19 20:36:54 | INFO | __main__ | Use Language: en_US
Traceback (most recent call last):
  File "C:\Users\Ryan\Downloads\Ilaria-RVC-Mainline-beta-2.0\infer-web.py", line 152, in <module>
    class ToolButton(gr.Button, gr.components.FormComponent):
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python311\Lib\site-packages\gradio\component_meta.py", line 198, in __new__
    create_or_modify_pyi(component_class, name, events)
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python311\Lib\site-packages\gradio\component_meta.py", line 92, in create_or_modify_pyi
    source_code = source_file.read_text()
                  ^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python311\Lib\pathlib.py", line 1059, in read_text
    return f.read()
           ^^^^^^^^
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python311\Lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 28981: character maps to <undefined>
Press any key to continue . . .

[proven hill]

the python to use is 3.10.2

[upbeat coral]

if I have both installed, and in my path and classpath how can I choose which it runs on

its going 311 by default

[proven hill]

uninstall 3.11

[ashen notch]

@viscid moss continue the same error.

[upbeat coral]

much easier than trying to manually change each module to match 311. I'll just do that and try thanks.

[viscid moss]

nah, that's not the index file.

The extension is literally .index

[fervent sun]

I got it now

[viscid moss]

Where did you extract the program?

[fervent sun]

it randomly popped up so i zipped it

[upbeat coral]

can ignore this right DEPRECATION: omegaconf 2.0.6 has a non-standard dependency specifier PyYAML>=5.1.*. pip 24.1 will enforce this behaviour change. A possible replacement is to upgrade to a newer version of omegaconf or contact the author to suggest that they release a version with a conforming dependency specifiers. Discussion can be found at https://github.com/pypa/pip/issues/12063 DEPRECATION: omegaconf 2.0.6 has a non-standard dependency specifier PyYAML>=5.1.*. pip 24.1 will enforce this behaviour change. A possible replacement is to upgrade to a newer version of omegaconf or contact the author to suggest that they release a version with a conforming dependency specifiers. Discussion can be found at https://github.com/pypa/pip/issues/12063

[fervent sun]

it doesn't specify, am i supposed to upload both the index and the model to hf?

[viscid moss]

yep, both files

[fervent sun]

also i got 2 index files, one says trained and one says added

and the added file is way larger

[viscid moss]

The right index is the ADDED

[fervent sun]

okay cool thank you

[viscid moss]

yw

[upbeat coral]

  File "C:\Users\Ryan\Downloads\Ilaria-RVC-Mainline-beta-2.0\infer\modules\vc\modules.py", line 195, in vc_single
    self.hubert_model = load_hubert(self.config)
  File "C:\Users\Ryan\Downloads\Ilaria-RVC-Mainline-beta-2.0\infer\modules\vc\utils.py", line 23, in load_hubert
    models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\fairseq\checkpoint_utils.py", line 423, in load_model_ensemble_and_task
    raise IOError("Model file not found: {}".format(filename))
OSError: Model file not found: assets/hubert/hubert_base.pt

Traceback (most recent call last):
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\gradio\routes.py", line 437, in run_predict
    output = await app.get_blocks().process_api(
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\gradio\blocks.py", line 1349, in process_api
    data = self.postprocess_data(fn_index, result["prediction"], state)
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\gradio\blocks.py", line 1283, in postprocess_data
    prediction_value = block.postprocess(prediction_value)
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\gradio\components.py", line 2586, in postprocess
    file_path = self.audio_to_temp_file(
  File "C:\Users\Ryan\AppData\Local\Programs\Python\Python310\lib\site-packages\gradio\components.py", line 360, in audio_to_temp_file
    temp_dir = Path(dir) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'
2024-04-19 21:02:44 | INFO | httpx | HTTP Request: POST http://localhost:7897/api/predict "HTTP/1.1 500 Internal Server Error"
2024-04-19 21:02:44 | INFO | httpx | HTTP Request: POST http://localhost:7897/reset "HTTP/1.1 200 OK"

@proven hill

[shell vector]

think you pinged the wrong person lmaoo

[upbeat coral]

ye i did and deleted it

[brittle wing]

Once I moved my voice model, since the gpu was running out on colab... to start the train again do I have to change google account to get the gpu again?

[upbeat coral]

i think version 3.34.0 gradio doesn't work which is why I am getting this issue

[queen escarp]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

https://docs.aihub.wtf/v/brazil

🇪🇸 Spain

https://docs.aihub.wtf/v/spain/

🇫🇷 France

https://docs.aihub.wtf/v/france

🇩🇪 Germany

https://docs.aihub.wtf/v/germany

🇯🇵 Japan

https://docs.aihub.wtf/v/japan

🇰🇷 Korea

https://docs.aihub.wtf/v/korea

🇵🇱 Poland

https://docs.aihub.wtf/v/poland

🇷🇺 Russia

https://docs.aihub.wtf/v/russia

🇮🇹 Italy

https://docs.aihub.wtf/v/italy

[quiet vale]

is there a way i can set the pitch to below -20 with this ui? https://github.com/Tiger14n/RVC-GUI

[mellow marsh]

theres any solution to use amd card for rvc?

[modern oxide]

На линуксе как-то можно вроде

[light root]

Which one is the best for ai rap covers?

[dawn nebula]

they are all using the same thing, it depends on the model

[light root]

Alright thanks

[mellow marsh]

есть ссылка?

[dawn nebula]

rocm

[sand iris]

lol good luck

[boreal mango]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface Spaces

• Applio, by IA Hispano Huggingface Spaces
• Ilaria RVC, by thestingerx Hugginface Spaces
• RVC-HFv2, by r3gm Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -local, -overtrain, -realtime, -rvc, -help

[torn osprey]

kinda struggling right now

im trying to set it up but when i drop the file

it says its not supported

[brittle wing]

yall what site to make ai songs

is the best

free

[torn osprey]

ngl im not too sure

my brain is about the size of a pea

it doesnt have a "." which is why im confused

i got it from the voicemodels channel but idk

[rare goblet]

Ayo? @torn osprey level 1 !!!

[torn osprey]

yeh