✨│ai-help

[half cove]

#1159513888199540817

[errant pond]

I hope it's a temporary issue btw. Or am I the only person experiencing it rn?

it basically has a trouble with downloading a file that's also from Huggingface and that actually exists?

[ember kraken]

Traceback (most recent call last):
File "/content/Retrieval-based-Voice-Conversion-WebUI/EasierGUI.py", line 371, in vc_multi
paths = [os.path.join(dir_path, name) for name in os.listdir(dir_path)]
FileNotFoundError: [Errno 2] No such file or directory: 'E:\codes\py39\test-20230416b\todo-songs'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/content/Retrieval-based-Voice-Conversion-WebUI/EasierGUI.py", line 376, in vc_multi
paths = [path.name for path in paths]
TypeError: 'NoneType' object is not iterable

How to fix this?

[eternal thunder]

hey there,

i am on epoch 56 now and want to ask for your opinion.

did my model allready overtrain?

[hot lagoon]

What is the d total looking like?

[jade pawn]

I've been trying to train my own celebrity AI singing voice, and no matter how many epocs I do (50 to 750), it doesn't come out sounding right. I've tried starting with different audio files for the training, but they all come out sounding the same, so I feel like the audio I'm starting with (3 different tracks from studio recordings and 1 longer live track) is either ok or it all has the same issue. I've tried shorter clips (1:30) up to almost 9 minutes of audio thinking that maybe it would sound better if I gave it more audio to work with, but they both sound the same too. I'm using The Mangio-RVC-Fork. Maybe that's not the best or most up to date program? Any tips are welcome.

[odd shale]

Um, your model graph is fine.

It hasn't overtrained still

[jaunty shale]

Is there a good google colab to make models on? I am still unsure if the other one is still banned or not.

[odd shale]

RVC Disconnected i think.

Other options are installing Applio or Mangio Fork locally if you have a good GPU and RAM

[jaunty shale]

Well.. I did try before, plus I asked for help like multiple times, but no good results

[odd shale]

No good results in what?

[jaunty shale]

Okay, I had this issue, whenever I trained a model with RMVPE or RMVPE-GPU locally, it did quite a quick process and it was done in about like what... 20 minutes?
But.. the problem was, the model itself had bad results, that it didn't even sound the exact same from the original dataset I put in

Like It sounded normal, no glitching, nothing. It was completely different voice out of it when I chose RMVPE

[rare goblet]

Ayo? @jaunty shale level 2 !!!

[jaunty shale]

Or more specifically, the steps had the same value as epochs, like as a result

[hot lagoon]

how to first get the background vocals and use those background vocals?

[jaunty shale]

i had this issue for months ever since RVC got shut down

[odd shale]

Download BVE model in UVR local under VR Arch download tab

Or use BVE V2 model in X Minus Pro (You'll must get an suscription to X Minus to do that)

[odd shale]

Maybe or your dataset is unproperly made, or your PC doesn't have the specs to support RVC

[jaunty shale]

1660 Ti, 16 GB RAM, i5-10300H

I will say though, it works on harvest, but it takes very long.

like I had to leave it for the whole night.

[odd shale]

I don't know if RVC supports 1660 Ti

[jaunty shale]

Heard it works for someone

[jaunty shale]

however, I haven't tried those before, I think?

[odd shale]

Try if Mangio Fork works for you.

Lemme get you an direct link download

[jaunty shale]

I used this one

[jaunty shale]

is there like a tutorial for it, or is it same?

[brittle wing]

how many epochs for a 10 minuite model?

[odd shale]

After downloading, just go to the Mangio root folder and double click on the go-web.bat file: https://huggingface.co/MangioRVC/Mangio-RVC-Huggingface/resolve/main/Mangio-RVC-v23.7.0_INFER_TRAIN.7z

There you have a direct download.

[odd shale]

Between 300, 400 or 500 depending on the tensorboard graph.

[brittle wing]

can you show me a photo of it?

[odd shale]

I don't have any.

It's matter of watching the graph.

Some 10 min datasets tend to reach their biggest low even before 300 epochs, so watch out.

[brittle wing]

nothing is showing on the tensorbord graph.....

[odd shale]

Which RVC version are you using?

And on which platform?

[brittle wing]

**Im using this colab **https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw?pli=1#scrollTo=G0MEhFM19Vq6

[boreal forum]

what site to use to download yt audio for the best quality

[brittle wing]

My colab is not detecting a gpu

anyone help me

i have a nividia gpu

hello?

anyone here?

[hot lagoon]

Can someon send me a colab to Applio google?

[jaunty shale]

does it work the same as the other one?

[odd shale]

Nope, Applio colabs were banned too.

[boreal forum]

is there anyway to train online?

[odd shale]

Just download it, it should work fine for inference and training

[jaunty shale]

do I need to download a tensorboard for this

like put the files in like in the other one

[odd shale]

I don't know.

[odd shale]

Only if you decide to install Mangio or Applio in Paperspace or Vast.ai

[boreal forum]

is there any tutorial for that?

[rare goblet]

Ayo? @boreal forum level 1 !!!

[jaunty shale]

found this tutorial https://www.youtube.com/watch?v=ixB9oalT3cQ&ab_channel=AIToolsSearch

maybe that could work

[odd shale]

That's a tutorial for using Mangio locally.

NOT online

[boreal forum]

what about this links?

[odd shale]

You can use those spaces too.

Even tho i suggest using Ilaria RVC space

[jaunty shale]

Yeah, I'm aware, I'm watching it right now

[boreal forum]

how to train on this model

[jaunty shale]

Although, there's one thing I'm thinking of rn..

How much you recommend for the numbers of CPU threads (I have 4 at the moment) and batch size?

[odd shale]

Those are Huggingface spaces.

[eternal thunder]

❤️

[rare goblet]

Ayo? @eternal thunder level 3 !!!

[eternal thunder]

so the blue part suggests overtraining. am i right? @odd shale @jaunty shale @hot lagoon

[odd shale]

But the lowest point was between 1.2k and 1.4k..

I don't know

Unless you're using an overtraining detector

[plain rivet]

-colab

[azure marsh]

All working colabs / spaces ☁️

• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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, -local, -overtrain, -realtime, -rvc, -help

[hot lagoon]

Did it just got ban or was this a long time ago?

[brittle wing]

im currently training a model

hopefully it turns out good

wish me luck guys

[proper shale]

it is quite likely it's overtraining, but check kl and mel

[odd shale]

Yeah, long time ago.

[proper shale]

oh and good luck

[harsh laurel]

Hi. Is there a working Applio-RVC-fork Kaggle notebook? The thing is that they must be send as a .ipynb files without public link so Kaggle won't detect and ban it

[proper shale]

Not any public ones afaik

[brittle wing]

how can i download this one?

[frozen dagger]

please help im mmvc on input i have vcable OUTPUT and on output i have INPUT

[scarlet cedar]

ISS Info

Location of the ISS now:
15.3846, -76.4250

Humans in Space (7):

**ISS**

1. Jasmin Moghbeli
2. Andreas Mogensen
3. Satoshi Furukawa
4. Konstantin Borisov
5. Oleg Kononenko
6. Nikolai Chub
7. Loral O'Hara

Nov 26, 2023, 5:48 PM

[candid prairie]

can any one please give me the link of both RVC v1 and v2

[proper shale]

#1159949278270193734, rvc v2 already supports v1 so you're finr

[gentle island]

Is there an easy way to disable some of the process the aicovergen space from huggingspace does? It tries separating instrumental and other stuff every time which I can already do elsewhere. A 19 second song cutout took like 700 seconds. Not sure how much the full song will take. Already past 1700 seconds

[abstract saffron]

If I rename the best backup of model to use it. Will it work ok?

[candid prairie]

hey if i want to clone my own voice and use it on my youtube videos does that will effect my youtube monetisation

[proper shale]

Sadly no

[proper shale]

Don't think so

[proper shale]

I suppose it will

Else just rename it back

[candid prairie]

oh!

[rare goblet]

Ayo? @candid prairie level 1 !!!

[abstract saffron]

I heard that youtube can off it if you use robot voice, and give reason:"Same content"

[proper shale]

Hmmm... I dunno if that's true

@candid prairie best bet would be just to try and see how your audience reacts to it

and how YouTube reacts to that too

[candid prairie]

ok i will try it, Thanks for help

[proper shale]

np

[gentle island]

Oof. Thank you for the answer. Can you only have one hugginspace space running on a free account? I tried launching AICoverGen earlier and it was throwing errors(I already duplicated a different space yesterday). I then deleted both spaces and duplicated AICoverGen again and it's working now. I was wondering if I should use the 'pause space' from settings if that's the case.

[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

[proper shale]

I think you can have more than one. I remember having both Ilaria RVC and RVC-HF on at once

So uh yeah it's possible

Honestly I'd recommend using the AICoverGen Colab in the pins, which was fixed to prevent those Colab restrictions

#✨│ai-help message

[gentle island]

Oh.....

I didn't know there was another version pinned here. Thanks

Also, is the last hugginspace space in #1159949278270193734 alright? The main one is showing errors

[rare goblet]

Ayo? @gentle island level 1 !!!

[abstract saffron]

Am I right that if I will train my model to 500 epohs, I just can use backup which was lower than other?

[atomic mulch]

What is the best way to remove background music from dialogue?

[abstract saffron]

This isn't the best but OK. https://colab.research.google.com/github/jarredou/MVSEP-MDX23-Colab_v2/blob/v2.3/MVSep-MDX23-Colab.ipynb#scrollTo=6C_pzkxrjzIr

[atomic mulch]

Is this more efficient than the local installed RVC?

[violet heron]

UVR

Locally since you get more options and it’s unlimited

[atomic mulch]

what is that?

[violet heron]

Ultimate Vocal Remover

It can remove background noises/music

About studio level quality

[atomic mulch]

is it apart of rvc?

[violet heron]

Not apart, different

[atomic mulch]

Would you use this or rvc local?

[violet heron]

But i do think that a few forks include it

[violet heron]

UVR is for removing background noises

RVC is for cloning voices and making covers

[abstract saffron]

It is if you have bad PC. It is same thing like Ultimate Vocal Remover.

[atomic mulch]

Thanks! where do you get UVR?

from here https://ultimatevocalremover.com ?

[violet heron]

Yep

[atomic mulch]

thanks a ton!

[summer perch]

Hi, I was going step by step to install aicovergen natively, but when I used the command "python src/download_models.py" it gave: module not found error: no module named "requests"

[violet heron]

Refresh

Click the arrow

And did you put the pth in weights folder?

[rare goblet]

Ayo? @left sentinel level 1 !!!

[violet heron]

Don’t know, try restarting it

[left sentinel]

restrated my whole comupter still deos not work

[atomic mulch]

Do you reccommend any settings for UVR?

[violet heron]

Not really

Set your output to the folder you want the outputs saved

And input being the audio file you want to separate

[atomic mulch]

This program is amazing thank you so much tim! @violet heron

[hot lagoon]

Have you try the BVE model in UVR

[violet heron]

No, why?

[hot lagoon]

Cuz the BVE gets you the backgourd vocals

[molten pecan]

Set the segment size at 320

The lower the more quality it has

iirc

[jaunty shale]

Trying mangio-grepe. it normally works like how it should, but training takes very long.. Any idea how to speed this up?

[atomic mulch]

the higher the number the more qualtity

[molten pecan]

Whaaaat have I been doing the other way around this whole time?

Is segment size the same as window size?

[modern oxide]

use rmvpe

[prime hemlock]

^

[jaunty shale]

I did

[rare goblet]

Ayo? @jaunty shale level 3 !!!

[jaunty shale]

I used the rmvpe, it's still slow

[modern oxide]

it's better than mangio-crepe

[jaunty shale]

Well I know that, the question is why it takes longer?

I even tried to use less datasets since I had them seperated

now it's like around 3-4 minutes long, 1 audio

[modern oxide]

what gpu do you have?

[jaunty shale]

1660 Ti

[modern oxide]

it's not the best gpu for rvc, so there's no way to make it faster than it is for you

you can upgrade your pc by buying a new nvidia gpu with more cuda cores and clock speed

or use colabs

[jaunty shale]

isn't colab dead?

or most likely disconnected?

[modern oxide]

no

[prime hemlock]

could be batch size

[modern oxide]

what batch size u usually put?

[jaunty shale]

Tried putting 4-8

[prime hemlock]

how much vram do you have

[modern oxide]

6 apparently

4 should be managable, but it has only 1500 cuda cores, so it still will be slow

[jaunty shale]

Translation:

1. Total memory:
2. Display Memory (VRAM):
3. Shared Memory

this is the GPU

[modern oxide]

6 vram and 16 ram

how much time does it take per epoch?

with batch size 4

[jaunty shale]

uuuh let me check hold on

Also, does this affect the speed/quality?

[modern oxide]

no, it just affects the speed of preprocessing and feature extraction

[jaunty shale]

Aah okay

[exotic pond]

Is it possible to download Kits.AI models for local use?

[rare goblet]

Ayo? @exotic pond level 1 !!!

[odd shale]

Nope.

[modern oxide]

how long is the dataset?

[jaunty shale]

4 minutes and 30 seconds

it was longer before, I just made it shorter

[modern oxide]

well, seems normal, you can't get faster speeds on your gpu

[jaunty shale]

What about the colabs? Are there any that can be used? That aren't banned?

[modern oxide]

i mean you can try to shorten the dataset to 1 minute, that will help the speed, but it'll also affect the quality

[jaunty shale]

yea that's true

[modern oxide]

yes, i presonally use easy colab

[jaunty shale]

may I see it?

[modern oxide]

you can find it in #1159949278270193734

[jaunty shale]

which one specifically is better? Or it doesn't matter?

[modern oxide]

i like easy colab more, because it's easy to use and automatically saves each epoch to your gdrive

rvc v2 disconnected has more settings u can change, but it doesn't automatically save your model to gdrive, so you'll have to wait till it finishes to export the model, or5 else it'll disconnect you and all the progress will be lost

[brittle wing]

when i try to convert the song it gives me a error can anyone fix it for me?

[jaunty shale]

where's the "easy colab" in #1159949278270193734 ? Cuz I am not sure which one it is. Apologies if that came off dumb from me

[modern oxide]

bruh, why is it not there

gimme sec i'll send you the link

[jaunty shale]

oh lmao

[modern oxide]

#1163883017446629508 message

here

easyGUI by rejekts

[jaunty shale]

alrighty

Imma try it

[jaunty shale]

this doesn't feel right.

[modern oxide]

did yo make sure that the dataset fully loaded?

it might take some time

[hot lagoon]

In what order do I use the BVE modle in uvr?

[jaunty shale]

wait, do I have to put it in the dataset as a zip file?

[rare goblet]

Ayo? @jaunty shale level 4 !!!

[jaunty shale]

wait--

[proper shale]

ooof

yeah it went wrong

[jaunty shale]

noticed

why though?

[modern oxide]

no, in easy colab you just put your audio

[rare goblet]

Ayo? @brittle wing level 1 !!!

[modern oxide]

and wait till it fully uploads

what happenned?

[proper shale]

explain your issue

just wait for a bit

[jaunty shale]

what if I have more than just 1, do I have to normally put it there?

[modern oxide]

yeah

[jaunty shale]

Oooooh I see what you mean now

[candid lichen]

@molten pecan I never did end up making it work lol

[molten pecan]

Oh 😭

Where did we left?

[candid lichen]

I tried 3 different models on 2 songs

For some reason when I tried to use a joji song it just kept on loading and loading it wouldn't stop lol

On the convert option

[molten pecan]

Have you tried doing it another web browser?

Like Brave or Firefox

I recommend Brave

[candid lichen]

I mean I have chrome and a ad blocker browser but guess I can try the Bing to see it that works if not I'll try Brave

[modern oxide]

yep

[molten pecan]

Brave is an app

[candid lichen]

Ik

[molten pecan]

Ah

[candid lichen]

So just to clarify it does work on mobile right

[molten pecan]

Yes

[candid lichen]

Okay

Hey so I noticed up in this chat that there's a vocal remover, is that free?

Cause I usually use vocalremover.org but I can only do it once

[molten pecan]

Yes absolutely

[molten pecan]

You have three unlimited and free alternatives

[jaunty shale]

works now

[molten pecan]

mvsep.com

Ultimate Vocal Remover 5

Or the huggingface space for UVR5

[hot lagoon]

In what order do I use the BVE modle in uvr?

[candid lichen]

Can I use that space?

[molten pecan]

Of course

[candid lichen]

You have the link for it or where do I go to duplicate it

[molten pecan]

Ah just google "ultimate vocal remover hugginface"

[candid lichen]

Okay gotcha, hey just wanna say thank you for helping me out I appreciate it alot

[brittle wing]

whats a dissconection prevention method

https://docs.aihub.wtf/guide-to-create-a-model/tensorboard-rvc#ix.-how-to-use-tensorboard does anyone know how do i get the rod file needed for tensorboard? When I click the download link it's not there anymore

[proper shale]

#✨│ai-help message

this installs TensorBoard

but not RoD

[hot lagoon]

[brittle wing]

thank youuu 👉 👈

[proper shale]

np

[flint cloud]

uhh can i use full youtube urls (biy that i mean with instrumental) on aicovergen

[candid lichen]

So any reason why I get a runtime error while trying to duplicate a space on huggingface?

[proper shale]

yesh

[flint cloud]

OH HELL YEAAAA

[candid lichen]

I repeat once more, why do I get a runtime error while duplicating a space on huggingface

[modern oxide]

no idea

[candid lichen]

Yeah and Google says it could be several different things so

What Does factory resetting a space do?

[rare goblet]

Ayo? @candid lichen level 4 !!!

[proper shale]

it could probably solve your problem

it'll remove any saved data, but that really isnt an issue

[candid lichen]

Well one thing Google mentioned saying it could be a GPU Thing?

[proper shale]

don't think it is

since only CPUs are free

[hot lagoon]

In what order do I use the BVE modle in uvr?

[proper shale]

@odd shale

help them out cuz u use BVE

[odd shale]

Okay Mr MJ.

Who's the one who needs help with UVR's BVE model usage?

[odd shale]

Let me tell you what to do.

1 - (Do this step incase you don't have the BVE model downloaded beforehand) - Download the BVE model under the VR Arch model download tab (after you selected the model, press on the big download button down below download sections)

2 - Wait till the model gets downloaded and select it via right click and select Advanced MDX Net settings

3 - Click on Vocal splitter options and check the bookmark that says "Enable vocal split mode" and select the BVE model. You can check the "Deverb vocals" checkmark in case you wanna get your lead or backing vocals dereverbed too.

[vital field]

what do i do @odd shale

[odd shale]

I don't train with Local Mangio, sorry but i'm unable to help you..

[hot lagoon]

How dose the Vocal splitter work tho?

[odd shale]

In Vocal Splitter options you can select various models as BVE, Karaoke 2, 5HP or 6HP to isolate your lead/backing vocals

But in order to vocal splitter to work, you must select any MDX model as Voc FT, Kim vocal 1 or 2, or MDX23C

It works like this (independently if you select an already isolated acapella or an unaltered song): First, the song will get it's vocals isolated, then it will start splitting the lead or backing vocals, and lastly, dereverb these vocals (incase you checked the "Deverb vocals" option)

You can select if you wanna get the Lead or Backings

[hot lagoon]

if I pick the one call All vocal typ will it do both?

[odd shale]

I'm not sure, but you can test.

Make sure to select an output folder.

[brittle wing]

How can I do inference ?

[hot lagoon]

Also when its done it ouputs like 10 flies Witch one to look for?

[violet heron]

What’s your GPU?

[odd shale]

the file that has "backing" on the name.

That's for backing vocals

the file that has "vocals_lead" stands for Lead vocals

[half cove]

you dont need gpu for inference

[brittle wing]

We’re can I find that all I see is that my Mac has a apple m2 chip

[sand iris]

Should be supported then

[brittle wing]

Wym

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

What do u mean

?

[sand iris]

You should be able to get some fairly fast inference on that

Keyword is should because macos support isn't tested all that much

[brittle wing]

Do u mean rvc local

Cause I can’t find any tutorials on how to use applio on MAC locally

[strong belfry]

You dont have a specific gpu because you have a apple m2 chip.

[brittle wing]

Ok

I was lost for a sec

Ty

[strong belfry]

If you download a rvc fork that runs on mac

you will be relying on the cpu

I believe applio does install on mac

not sure if it will just run

[brittle wing]

Where can I get both on the links to it

[strong belfry]

-rvc

[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

[strong belfry]

-fork

uh

-applio

[azure marsh]

All Working Local Forks 🖥️

• Applio-RVC-Fork, 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 Huggingface

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

[strong belfry]

here

[brittle wing]

Ty

Is there a tutorial on how to run it ?

Hello, is there any tutorial on how to download RVC locally?

[proper shale]

all in one documentation in #1159513888199540817

[brittle wing]

Was originally using app.kits.ai but I'm migrating to local since I've ran out of free minutes for conversion, plus only 2 models allowed is pretty bad.

So yeah, am planning on moving on to local.

Thank you!

[proper shale]

yayyy

[proper shale]

np

[brittle wing]

How good does RVC handle low training data count? I liked Kit because I managed to get decent results when training on a character who has only 1 second of total dialog avaliable, so it's very versatile.

[proper shale]

kits = rvc

[brittle wing]

Okay, good.

[proper shale]

lol

so ur gonna get even better quality actually

because u can change pitch extraction n stuff

[brittle wing]

Probably, since I'm not paying them a penny they only let me train low-fidelity models.

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

Which makes the output sound kinda like it was recorded on a bad mic.

[proper shale]

i mean, short datasets are actually a bit meh

so idk if it'll be thaaaat good

[brittle wing]

Yeah, I'm only using short datasets because they're the only avaliable count.

So, essentially for Kits, you pay them to do the same thing you can do locally? And not only that, they just completely got rid of that character hub where you can find people's models?

That's very scummy.

[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

[proper shale]

yeah

but i mean

they're limiting every aspect

I'd understand if it was like, some aspects

But it's everything

They locked down everything

Thing is, for doing local stuff, you do need a good GPU, and for companies that costs a bit especially if you're buying so many GPUs to use at once

[brittle wing]

I wish I could download my Kits models, my brother probably would want to do that too instead of retraining the Azur Lane models he made after I showed him that site.

[brittle wing]

I mean, you can also use google colab, so even someone on phone could probably make A.I. covers.

[proper shale]

Yes that's what I do haha

[brittle wing]

I'm gonna run locally though.

[proper shale]

It's the best option

[brittle wing]

Could probably speed up training a bit, dunno how good google colab's gpus are.

[proper shale]

T4 is pretty okay

But I think RTX 30+ series are already faster

[brittle wing]

I have 8 GB of VRAM in my GPU, and if they let you offload into RAM, I have 24gb of RAM.

[proper shale]

ooh nice 8gb is enough

[brittle wing]

Yeah, figured. I can run language models locally with that GPU. (Just planning on upgrading to a better GPU to run 70b models)

[odd shale]

Paperspace GPUs are good too.

A suscription right there is cheaper than Kits suscriptions.

[proper shale]

Paperspace is a cloud service that you have to pay tho

¯_(ツ)_/¯

I'd rather do everything locally

[brittle wing]

Well, if I can run it locally, I'll always pick that over any paid option.

[odd shale]

For that reason i've said its more reliable and CHEAPER

[brittle wing]

Much cheaper to just be local though, $0.00 lol

[proper shale]

electricity bill tho 💀

[odd shale]

Tho ☠️

Lol

[proper shale]

But it's probably cheaper

Than cloud soluy

solutions*

[odd shale]

I'm happy enough with Paperspace because my laptop has no GPU

It only has an Intel Iris Xe Graphics

[proper shale]

my laptop has Intel HD Graphics 4000

so you're lucky haha

[brittle wing]

If you have a good enough CPU you could probably run it CPU only.

[rare goblet]

Ayo? @brittle wing level 2 !!!

[brittle wing]

Would just take ages.

[proper shale]

Yep

Thing is, I got a 3rd gen dual-core i7.

soooooo

[odd shale]

My CPU is an Intel Core I5

[brittle wing]

Clicked on the original RVC thing on huggingface, now it's just downloading some file, what's this anyways?

[odd shale]

And 16 GB RAM

[proper shale]

probably the install bat file or the zip?

[brittle wing]

I mean, wouldn't using CPU only use the same amount of RAM as it would in GPU VRAM?

[brittle wing]

it's a 7z file, dunno what's in it though

and it's 5gb

[proper shale]

oh everything should be in it already then

[brittle wing]

Oh okay good then

[proper shale]

not really, since GPU has its dedicated VRAM

[brittle wing]

Kinda awkwarwd that it isn't just some kind of download page and instead a direct download

[proper shale]

CPU just uses whatever RAM is in the system

[brittle wing]

ohhh i see

yeah i'm not really good with computers

for the most part i get my friend to help me on ai adventures

[proper shale]

It's ok lol

Can't say I'm that good at PCs either (I have no clue how to build one with all the parts haha)

[brittle wing]

Yeah for my second gpu i'm just gonna get my brother-in-law to help installing it lol

[proper shale]

Oh nice

[odd shale]

🐢

[brittle wing]

hmm

does aicovergen have functionality for training?

[odd shale]

No.

[brittle wing]

ah damn

would be cool since

webui

[proper shale]

Nope, sadly

[odd shale]

It's an Colab for AI cover making only.

[brittle wing]

granted you can also download it locally

[proper shale]

...but honestly

I don't think it needed a training model function

yeah

[odd shale]

A room with AI generated people.

[brittle wing]

well then aside from original rvc which one does have a training functionality?

[odd shale]

Mangio Fork and Applio

Applio is basicly an Mangio with more features and some optimizations

[brittle wing]

Then I'll choose applio.

it being installed by a bat file is also sending good signals to me

instead of just a 5gb zip file lol

[glad zealot]

Mangio doesn't work on colab.. I've tried making that thing work

Applio works tho

[brittle wing]

well i'm not using colab personally

[odd shale]

Thats kinda hilarious since both are RVC forks.

[odd shale]

Me neither

[glad zealot]

Ye I have no idea why mangio or mainline doesn't work

What kind of sorcery does applio have that it works on free tier

[brittle wing]

i remember i first used colab for pygmalion

before i realized pygmalion is a very MEH model and switched to better ones locally

[odd shale]

I don't know, ask Blaise or Deivih.

[glad zealot]

Kek should I just dm? XD

[plucky locust]

arianatalking is not a registered model! Use /list to list all models, or use /register to register a new model.

[glad zealot]

Dm blaise

[odd shale]

Yeah XDD

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

[glad zealot]

Aight wish me luck

[odd shale]

Ariana Chiquita.

[glad zealot]

Only thing I can think of is gradio version

Wait no wut

Mainline is using 3.14.0 then both mangio and applio is using 3.34.0

Wtf??

[plucky locust]

Sorry! You need to register an account to continue making free covers. Please use the /login command.

[thick summit]

why does rvc take so long

[outer isle]

Hey

[glad zealot]

Cause it needs to compute stuff

[hot lagoon]

is it me or when geting a modle in Paperspace to the web the modle pth is not there?

[thick summit]

i know

[brittle wing]

then why ask that question?

[thick summit]

but ive used this before and it didn't take so much

but it does now

like almost an hour for just an audio of 3 minutes

[glad zealot]

No idea...never tried it

@proper shale ya used paperspace before?

[plucky locust]

Sorry! You need to register an account to continue making free covers. Please use the /login command.

Sending you a private login link

[harsh scarab]

You're placed in queue. Current position: 1

[toxic topaz]

Did Did you find a solution?
I have the same problem (RUNTIME ERROR), no matter which Space I clone. Factory reset didn't solve anything btw

[plucky locust]

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

[half cove]

@outer isle #1159290666019471421

[outer isle]

Why!?

[half cove]

So it doesn't clog chat

[modest topaz]

help i just tried downloading mangio RVC and when i run go-web.bat it doesnt work

[candid lichen]

No I didn't unfortunately

[molten pecan]

Did you try using a different model?

Or in another web browser?

Maybe the problem is either the browser or the model

[toxic topaz]

Yes. IIaria, Aesthetic and the uusuual one

[toxic topaz]

Firefox and Edge, yes.

[molten pecan]

When I say model I mean voice model

Not space

[toxic topaz]

I don't even get to that point. Before I even try to run the space, I get the Runtime Error error.

[rare goblet]

Ayo? @toxic topaz level 1 !!!

[odd shale]

Mija, creo que deberías aplicar para Helper.

[toxic topaz]

What happened?

[molten pecan]

Lo consideré. Pero ni siquiera soy una trusted member 😭

[rare goblet]

Vote Booster: 10% (4 hours remaining)
✨ Tip: Use /card to modify your rank card

[toxic topaz]

Alv podiamos hablar español kksjsjsj

[molten pecan]

Top 13 y aún no tengo perms

😦

[odd shale]

Yo no soy Trusted Member, y aún así me aceptaron.

[molten pecan]

Ah. Que raro

[molten pecan]

#🌍│español

[odd shale]

Solo intenta poniendo el comando /jointeam

[molten pecan]

Cuál es tu forma de duplicar los espacios?

[odd shale]

Pues, no estás haciendo nada mal.

[toxic topaz]

Es lo que no entiendo

Lo unico que queda es hacerme otra cuenta en hugginspace

[odd shale]

Intenta reiniciar el espacio.

[toxic topaz]

Hice eso y el factory reset

[odd shale]

Prueba con otra cuenta.

[candid lichen]

Yeah even when I factory reset it,it didn't do anything

[toxic topaz]

gonna try that

[odd shale]

Y, les recomiendo que vayan al canal de español.

No quiero que a alguno un Mod los regañe.

[modest topaz]

??

[toxic topaz]

No problem. I'm comfortable in English. Sorry about that xd

[odd shale]

No, lo decía por seguridad.

[toxic topaz]

Thanks for the help anyway 😄

[odd shale]

De nada bro.

[candid lichen]

I seriously wonder if it has something to do with the GPU

[toxic topaz]

Free Hugginspaces are CPU only, that's why they are slower than in Colab.

That's what I understand, at least.

[candid lichen]

Wonder why a conversion took like 8 minutes lol

[glad zealot]

Ya pretty much

[candid lichen]

Also when I did a conversion it loaded the export as html

Instead of wav or mp3

What do I do about that error?

[toxic topaz]

welp

A completely new account, duplicating a repository from here

In incognito mode. I guess it's something wrong with my pc somehow.

[ripe talon]

so uh. when I try to create a model. the index doesn't wanna appear

on my recent one

if this is the right channel.

for help

[molten pecan]

Listo ya subí mi currículum

:3

[molten pecan]

What error?

[odd shale]

Ojalá te lo acepten, porque serías una gran adición amiguita

[molten pecan]

Gracias :3

[odd shale]

De nada amiga.

[candid lichen]

The error in my sc

[tulip cloak]

Hello does interference work on MacBook

[molten pecan]

Those are the logs. Not an error message

It's just showing the stuff that's loading up

It will take a few minutes

[candid lichen]

I've done it 2 times and each time I've waited it just says runtime error

Should I just leave my phone on that screen and not click off?

[molten pecan]

Yup

[candid lichen]

So would that explain it then or?

[molten pecan]

Possibly

Runtime error seems to be a connection error

[candid lichen]

I'll leave it and see if it's different

[molten pecan]

So don't tap out and make sure your wifi is stable

[tulip cloak]

-colab

[azure marsh]

All working colabs / spaces ☁️

• AICoverGen-NoWebUI, by Ardha Google Colab
• RVC Disconnected, Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab
• RVC-HF, by r3gf Huggingface Spaces
• AICoverGen, by r3gf Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces
• Advanced RVC Inference, by neuclya Huggingface Spaces

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, -local, -overtrain, -realtime, -rvc, -help

[molten pecan]

@candid lichen did it load?

[candid lichen]

It was running and downloading but then this just popped up

[molten pecan]

That's so weird

Maybe try using WiFi instead of mobile data

Or try it in a computer

[zinc nymph]

I just had a quick question idk if anyone would be able to help, im attempting to use the merge lab and mess with some of the sliders but when I click merge nothing happens. Lines of code go through my console, but nothing changes in the actual application, nothing appears in my final slot or anything. Is a file made or something?

[glad zealot]

Go to your model_dir folder and look for a folder called 200

It's bugged atm it's supposed to be on 199 but it goes to 200 instead

[candid lichen]

Too bad I can't really use laptop so I guess I'll just have to use something else

But I'll try wifi

Errors just keep going lol

I give up ill just use a app or something lol

[molten pecan]

This is so weird 😭

Try using it on desktop or make a new acc

That's the last resource

[glad zealot]

Why not try the mdx colab one!

[candid lichen]

I can try but I don't really know what I'm doing

[molten pecan]

A bit cluttered for mobile and a bit hard to set up

[rare goblet]

Ayo? @molten pecan level 50 !!!

[molten pecan]

Not really hard, just not as straight to the point as huggingface

But sure they could try

[glad zealot]

Better than not working at all

[molten pecan]

I don't understand why that happens

Guess AI technology is still primitive

😭

[candid lichen]

I'm guessing it's because of me being on mobile I would assume

[glad zealot]

Why sure why being on mobile does anything since it's running on a VM on the website not your phone

Not sure*

[molten pecan]

Exactly

[glad zealot]

Autocorrect doing autocorrect stuff

[molten pecan]

Skill issue

[candid lichen]

That's what I would Like to think but idek

Probably just better if I have a separator application on my laptop anyways

[toxic topaz]

@candid lichen You and I seem to be the only ones with this problem. I would love to know what we have in common, but from what I see our enviroments couldn't be more different. I'm on WiFi from my pc and used several browser, for example. It is so weird.

Are you also from latam, by any chance? Maybe its a server issue idk

[molten pecan]

Probaste en una cuenta nueva?

[candid lichen]

I can see if making a different account does anything different

[toxic topaz]

@molten pecan yes.

[rare goblet]

Ayo? @toxic topaz level 2 !!!

[toxic topaz]

New account. No changw

[molten pecan]

Usaste diferentes motores de búsqueda o diferentes navegadores de Internet?

[toxic topaz]

Edge and Firefox.

On Windows

[molten pecan]

Usaste diferentes modelos?

A lo mejor el problema es el modelo de voz

[toxic topaz]

It was working until a couple of days ago. I never updated my pc, and other projects (not on my account, but public) work fine.

[toxic topaz]

Si, ya lo habías preguntado. Lamentablemente no llego hasta ese punto. Al iniciar el space da el error

Probé otros spaces de inferencia. Es siempre igual

[molten pecan]

Que raro

Si es un problema de servidor entonces usa un VPN

Que no esté conectado a latam

[hidden laurel]

can someone help me

when you are training a model , which is the best thingy to put it on

rmvme?

[molten pecan]

rmvpe

Either mangio-crepe or rmvpe. rmvpe should be your go-to algorithm

Try and see which one sounds best

[candid lichen]

So I think it's something with that link

[rare goblet]

Ayo? @candid lichen level 5 !!!

[molten pecan]

Congrats you can now talk in #✦│chat

[candid lichen]

I made a 2nd account and still did the same thing

Weird stuff

[hidden laurel]

thank you

[toxic topaz]

Yeah, im out of ideas honestly. Gonna try on another pc or just wait it fixes itself somehow

I appreciate all the help people have given me here though. Thank you very much everyone

[jaunty shale]

@modern oxide
Tried the results, it works!

Been trying to figure this out for months, finally a good solution. Thank you so much!

[inland canopy]

why does it say i need tensorboard? im using an amd card btw, trying to install applio

[molten pecan]

Maybe use a VPN

Like I said

[inland canopy]

@median smelt

[rare goblet]

Ayo? @inland canopy level 1 !!!

[inland canopy]

??

i really need help..

lmk

ping me lol

[median smelt]

:/

[median smelt]

can u help blitzs

[molten pecan]

I'm going to bed now sorry

[median smelt]

damn

alr

[brittle wing]

Try installing it at root directory

D:\ or C:\

Yet to see if this works but thats what was reccomended to me when i had this

[lofty fulcrum]

What do you do with the link you get from the aicovergen Google collab

[glad zealot]

You paste it on your browser

[gaunt kestrel]

for applio, im missing the VR models for uvr5

[rare goblet]

Ayo? @gaunt kestrel level 1 !!!

[gaunt kestrel]

does anyone know a fix

[lofty fulcrum]

It doesn’t work

[rare goblet]

Ayo? @lofty fulcrum level 1 !!!

[glad zealot]

Try running it again. Also you're using the modified version right?

[lofty fulcrum]

Idk it starts with a slash and ends with .mp3

[glad zealot]

I meant the link for covergen

[brittle wing]

What about harvest?

[lofty fulcrum]

Oh I just did a Google collab thing and I when through it and it said /content/AICoverGen/song_output/Bn3hZjjFTJA/Ken Carson - Jennifer’s Body (vocals only) (hev Ver).mp3 as where to get the song (please don’t criticize me for what I’m trying to do lol)

[glad zealot]

I meant the link on your url bar on google colab what is it

[lofty fulcrum]

Ok

[candid lichen]

Hey guys so after you hit convert is it suppose to go past 1000?

Hello

[finite galleon]

My ...

[candid lichen]

It's a yg song

[finite galleon]

Ahhh

[candid lichen]

But yeah anyways those numbers keep going up

[finite galleon]

For me it once got to 1200

[candid lichen]

It's almost at 2000

[finite galleon]

I mean your using a cpu so

Yeah it should be slow as heck

[candid lichen]

Well it's just weird cause when I used a James hatfield model it loaded up around 700

[finite galleon]

Use this https://colab.research.google.com/drive/1WMFpF_1e639kyR81KCd0ODQeLs26VOBt

Colab uses a gpu and inferring takes like 1 minute (to load scripts) and then its just 30 seconds

[candid lichen]

What do I put in the audio url thing

[finite galleon]

Skip it

Thats optional

The only thing you need to use is install downlosd model from url and inference

[median isle]

Every time I speak the res number goes really high and the voice comes out like o- o- o a- a- a can anyone help with that

[finite galleon]

And on inference just put in the model name and the song file path

[jade ginkgo]

Can someone please make me an AI cover? I cant do it rn because Ive been trying since yesterday and it hasn’t been working.

[rare goblet]

Ayo? @jade ginkgo level 2 !!!

[finite galleon]

The everything else is optional

[jade ginkgo]

Like the waiting time for the result would be like 2000 hours and then it just breaks

[finite galleon]

HuggingFace?

[jade ginkgo]

Ilaria rvc

I was wondering if I could ask someone to do it for me, because when Im on mobile, I cant sit there and wait and stay on the same tab doing nothing while waiting. So I was wondering if someone else could do it for me since this is getting really frustrating

[slate latch]

yo

why doesnt it work on the website when i try to upload the voice model

[finite galleon]

Use colab

[candid lichen]

So in the download models using irl what do I put

[median isle]

I'm very stuck

[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

[jade ginkgo]

Isnt that google colab?

Thats what was linked in the tutorial

[modern oxide]

No problem

[finite galleon]

Ofc the rvc model url

[candid lichen]

I just used something else to make it

[median smelt]

tried that, doesnt work

[brittle wing]

hm

[rare goblet]

Ayo? @brittle wing level 3 !!!

[brittle wing]

try c:?

c:\

[median smelt]

yes

i did

[brittle wing]

what about d:\

[median smelt]

but it doesnt work :/

[brittle wing]

because installing at c:\ fixed it for me

[median smelt]

wai

hmm

uh

like i put my tensorboard in there or what

[brittle wing]

no

you don't put tensorboard there

you install RVC at c:\

delete rvc and reinstall it at root directory

[median smelt]

kk

[rare goblet]

Ayo? @median smelt level 1 !!!

[red kayak]

The performance of ur phone is not affected whatsoever

Ur just running a webpage

[candid lichen]

I just think it's the link itself

[red kayak]

What about it

[candid lichen]

It just keeps getting errors

[red kayak]

Is it a webui?

[candid lichen]

Yeah for UVR On huggingface, so instead I just downloaded the program on my laptop instead

[red kayak]

What are ur laptop specs

[candid lichen]

Pentium silver, 4gb ram

[red kayak]

Yeah no

[candid lichen]

It did a conversion though

[red kayak]

Ditch the idea

[red kayak]

Yeah it's slow no

?

[candid lichen]

It took about ten minutes for the vocals and instrumental

[red kayak]

Try using mvsep

[candid lichen]

And wait in a queue?

I'm good

[red kayak]

If u have an account it shouldn't take long

[candid lichen]

I'll try it out next time

[inner mica]

Hi there I just wanted to know how to make a voice model of my own voice ?

[hot lagoon]

I got this whne opeing uvr
Traceback (most recent call last):
File "UVR.py", line 288, in <module>
File "UVR.py", line 181, in load_data
_pickle.UnpicklingError: invalid load key, '\x00'.

[near pasture]

were you able to solve this?

[candid prairie]

Hey I have tried to create an AI cover with the japanese song on youtube nut it is taking too much time to process

[slate latch]

How do I put another voice model in the website version?

[modern oxide]

there's no one click training button in mangio

only in og rvc or applio

[small zodiac]

yo how tf do u like

chnage the pitch

in huggingface RVC

[modern oxide]

there's a setting called transpoze

[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

[onyx rover]

You know here is hina mod aicovergen but I see white screen instead of gradio interface

[jovial sequoia]

(F0 det) what do you guys use in this setting?

[quick briar]

can I create my own model without using Google Colab? I can't find a tutorial that doesn't use it and I just wanna do it locally with RVC

[onyx rover]

It will take the long time to train model on standart gpu

[quick briar]

that's fine

[onyx rover]

I just see nothing when join

[modern oxide]

what are you trying to use?

[onyx rover]

hina mod

[modern oxide]

is it an rvc of some kind?

in colab?

[onyx rover]

<https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb this

[modern oxide]

oh

sadly gradio user interface was disabled for free users in colab

[glad zealot]

Seems to work for me

[glad zealot]

No it works

[reef snow]

hi

i neeed help

[onyx rover]

Why don't I see any object?

[modern oxide]

wut

[glad zealot]

Might be your internet connection or try switching to a different browser

[modern oxide]

it doesn't for me

and it shouldn't

[glad zealot]

It just loaded for me

[modern oxide]

because gradio was banned for free users

in colab

[glad zealot]

From the link they sent

[onyx rover]

What browser do you use?

[reef snow]

hello

[glad zealot]

Naaa

[glad zealot]

I use edge on PC and via on phone

[onyx rover]

I use too

I tried Ngrok but it just give some error when launch

[modern oxide]

yes, that is true, or else we wouldn't have created rvc v2 disconnected colab and easy colab

[glad zealot]

I wouldn't have made the modded versions if they don't work

[modern oxide]

well it doesn't open for me or for him

[glad zealot]

Ya that's what's weird

Location maybe?

[onyx rover]

I connected my laptop to the internet cable, no result

[glad zealot]

Def not internet connection since I'm on my mobile data

Try rerunning it

Disconnect and delete runtime

[onyx rover]

I used vpn, and it worked! Thanks

[rare goblet]

Ayo? @onyx rover level 2 !!!

[modern oxide]

weird, how did you make gradio work in colab? it was banned like 2 months ago in colab and it was a catastrophy

or maybe only rvc colabs were banned

[glad zealot]

Rot13 pretty much

The only gradio atm that I can't make work is mangio/tweaks and mainline rvc

[modern oxide]

hmm

aight

good to know

[onyx rover]

Ok, thank you

[glad zealot]

Applio works but meh imma let applio devs figure how that works

[finite galleon]

The applio colab bypass got junked

[glad zealot]

Wdym

[finite galleon]

It was working a week ago and it was completely broken now

[finite galleon]

There was an applio bypass

[glad zealot]

Lmao their version

[finite galleon]

Its pretty unknown

[glad zealot]

Mine still works

You want it?

[finite galleon]

Yeah because the colab got junked by the colab team

[finite galleon]

Nah i am good

[onyx rover]

@glad zealot, may you give links for all you works?

please)

[finite galleon]

Give it to the other people here tho

[reef snow]

i have problem with installation

[glad zealot]

https://linktr.ee/_hina__

[finite galleon]

They litteraly cant handle a single word on a colab that isnt a web ui

[onyx rover]

thank you

[reef snow]

hello sir

[rare goblet]

Ayo? @reef snow level 1 !!!

[brittle wing]

hello

i need help

i am trying to do the vocal separation but i get the following error

\AppData\Local\Temp\gradio\d5234f436a4078ca3e66868c38a3366ad2d4d9ce\13: No such file or directory

[modern oxide]

well, it's not present anymore, i dunno why

maybe devs thought that it's unnecessary

[upbeat talon]

Hello everyone, does anyone know a service for training a voice model online?

[brittle wing]

?

[rare goblet]

Ayo? @brittle wing level 7 !!!

[brittle wing]

-_-

@viscid moss sorry for ping but i dont really have much time to mess with this to figure it out myself

[viscid moss]

What's the problem?

[brittle wing]

it basically aborts right here

im not sure whats wrong but it isnt liking the links im giving it apparently

the one thats in it by default seems to work

yup worked

my own links didnt work

[viscid moss]

interesting......
Try with ur link and show me the output message

is an age restricted video?

[brittle wing]

uhm

i did

[brittle wing]

and this is what happened lol

[brittle wing]

i dont think so

[viscid moss]

[brittle wing]

am i being really dumb lmao

[viscid moss]

This can be annoying but try the audio through Drive

And.... can u share me that yt link?

I wanna investigate what's happening

[brittle wing]

literally just this https://youtu.be/S6G3e_C4e7s

not age restricted

[viscid moss]

Ok, I will test

Try downloading and mounting drive

[brittle wing]

alright

[viscid moss]

The optional step

[brittle wing]

seems to work

[viscid moss]

Interesting....

[brittle wing]

i have now realized that my model is shit and will retrain lmao

[viscid moss]

Noooo

[static gyro]

hello what do i do when the voice sounds like auto tune?

[rare goblet]

Ayo? @grand junco level 1 !!!

[viscid moss]

This was the problem, that shitty record label has copyright

I'll add that to the guide, ty for the advice

[glad zealot]

Ohh

[viscid moss]

Hmmmm that's difficult, I just remove the background sounds and leave the voice with autotune

[brittle wing]

oh interesting lmao

[glad zealot]

First time I've seen that

[viscid moss]

2nd for me

[copper pebble]

LMFAO

[viscid moss]

Age restricted videos doesn't works too

[upbeat talon]

How to make voice model with
RVCHFpriv(where s take the tutorial or who can explain?)

[viscid moss]

for that reason I made the GDrive support for that colab

[viscid moss]

U can't make (train) a model in Hugging Face RVC

[upbeat talon]

• i dont know how to do it

[copper pebble]

just use colab

#1159513888199540817

u should finde the rvc disconnected guide in there @upbeat talon

[brittle wing]

is training supposed to be this fast?

8 min dataset

[copper pebble]

yh

[brittle wing]

lmao

interesting

i guess training 8 minutes is faster than training 50 minutes 🤯

[copper pebble]

ur actually training 1 epoch in 14s

[copper pebble]

you can see it right here on the right

[grand junco]

i trained a model in applio rvc and pressed Train feature index button but it doesn't show any output messages and no index file is saved

[brittle wing]

i know

[viscid moss]

OMG Applio issue

[copper pebble]

lmao eddy

[brittle wing]

applios kinda mid compared to disconnected

imo

[viscid moss]

How long is ur dataset? Too short datasets 've problems @grand junco

[viscid moss]

RTX 4090, 30 min of dataset. This is fire 🔥

[brittle wing]

💀

stop flexing

im on an rtx 3060 12 gb lmao

[grand junco]

i think dataset is under 5 minutes so maybe that's a problem (i know dataset should be like at least 10 minutes but i was experimenting for no reason)

[fallen vale]

i want master wu rvc to read my text

what shoudl i do

please help

[viscid moss]

5 minutes should be fine to generate an index. I was talking about a few seconds of dataset.

This can be a bullshit but... u can try make the index file in Colab

if u really want the model, I know it's annoying

[viscid moss]

use TTS

#1159513888199540817

[plucky locust]

#1178186208996765716 message is not a registered model! Use /list to list all models, or use /register to register a new model.

[vapid gust]

what’s the shortest amount of time in ur dataset could work

[brittle wing]

does this look like overtraining to any of yall

[copper pebble]

yh i got really good models even with seconds of dataset

[brittle wing]

5-10 minutes is optimal

[copper pebble]

did even 20s

[brittle wing]

yeah

[vapid gust]

I just asked the shortest amount of time

[copper pebble]

it really depends on your dataset quality

[vapid gust]

what if it’s studio quality

[upbeat talon]

The collab has collapsed

[vapid gust]

like straight out of the studio

[copper pebble]

??

[grand junco]

can i just use og rvc locally for making index instead of colab?

[vapid gust]

Was it good?

[copper pebble]

yeah i guess

[copper pebble]

not bad at all

[brittle wing]

uh ok so do i stop it then lmao

[copper pebble]

but

try with 5 min

@vapid gust

[vapid gust]

nah cuz I got someone rapping studio vocals

but its like 1.3mins

[upbeat talon]

I'm a novice in this business and need specific instructions

[vapid gust]

I just want to know if it will come out good

[brittle wing]

wait actually i think this is fine

[karmic locust]

would it be okay to post a kind of long error here for debugging?

[viscid moss]

https://docs.google.com/document/d/168uFdelqp6Mwx68VIRywWxqi4LHTmziG0jUZ8-1de2g/edit?usp=sharing
C'mon bro, u can do it

[copper pebble]

yh thats good

[copper pebble]

dm me im gonna follow you trough this

[karmic locust]

I dont know if this is the correct place to ask this really, i had all this stuff working hours ago, idk what i did that broke it all but now whenever i run the "go-realtime-gui.bat" it gives me an xformers error every single time and crashes out:

C:\AI_Stuff\Audio\RVC1006Nvidia>runtime\python.exe gui_v1.py
2023-11-27 09:05:20 | WARNING | xformers | WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:
    PyTorch 2.0.0+cu118 with CUDA 1108 (you have 2.1.1+cpu)
    Python  3.9.13 (you have 3.9.13)
  Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)
  Memory-efficient attention, SwiGLU, sparse and more won't be available.
  Set XFORMERS_MORE_DETAILS=1 for more details
2023-11-27 09:05:21 | INFO | faiss.loader | Loading faiss with AVX2 support.
2023-11-27 09:05:21 | INFO | faiss.loader | Successfully loaded faiss with AVX2 support.
Traceback (most recent call last):
  File "__init__.pxd", line 942, in numpy.import_array
RuntimeError: module compiled against API version 0x10 but this version of numpy is 0xf

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\AI_Stuff\Audio\RVC1006Nvidia\gui_v1.py", line 68, in <module>
    import tools.rvc_for_realtime as rvc_for_realtime
  File "C:\AI_Stuff\Audio\RVC1006Nvidia\tools\rvc_for_realtime.py", line 13, in <module>
    import pyworld
  File "C:\AI_Stuff\Audio\RVC1006Nvidia\runtime\lib\site-packages\pyworld\__init__.py", line 7, in <module>
    from .pyworld import *
  File "pyworld\pyworld.pyx", line 6, in init pyworld.pyworld
  File "__init__.pxd", line 944, in numpy.import_array
ImportError: numpy.core.multiarray failed to import

I am using the RVC1006Nvidia release linked on github, again, this all worked fine and flawlessly hours ago 😞

[rare goblet]

Ayo? @karmic locust level 1 !!!

[vapid gust]

ai snm

cuz I got a lot of rappers but there verses r short yk

n maybe 2 songs but 30-45sec verses

[copper pebble]

if the max u can do just try

[karmic locust]

also it keeps saying i have python 3.9 in this error but i have uninstalled every python and reinstalled many times over now to make it see the correct one and it won't

[copper pebble]

training??

[karmic locust]

no, the realtime voice conversion

is this the wrong place for this report?

[molten pecan]

Mhm

[copper pebble]

cd venv/Scripts
activate
pip install --force-reinstall torch torchvision --index-url https://download.pytorch.org/whl/cu118
pip install --force-reinstall --no-deps --pre xformers

[molten pecan]

This channel is for RVC

[copper pebble]

try this code

[molten pecan]

For the realtime voice conversion go to #🔍│help-w-okada

[karmic locust]

but the thing im using is not okada

[copper pebble]

it says : Please reinstall xformers

[molten pecan]

What are you using?

[karmic locust]

yeah i have tried it but i will try it again

[copper pebble]

did you get that from our server??

[candid prairie]

can anybody tell me any free TTS software like elevenlabs which can help me

[copper pebble]

the vc ??

[copper pebble]

11labs or ilaria tts

[karmic locust]

"Retrieval-based-Voice-Conversion-WebUI"

is this discord link to that?

[radiant flare]

local rvc w. nvidia / linux ubu 22.04

<All keys matched successfully>
Traceback (most recent call last):
  File "/home/default/Downloads/RVC0813Nvidia/lib/audio.py", line 14, in load_audio
    ffmpeg.input(file, threads=0)
  File "/home/default/.local/lib/python3.10/site-packages/ffmpeg/_run.py", line 325, in run
    raise Error('ffmpeg', out, err)
ffmpeg._run.Error: ffmpeg error (see stderr output for detail)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/default/Downloads/RVC0813Nvidia/infer-web.py", line 188, in vc_single
    audio = load_audio(input_audio_path, 16000)
  File "/home/default/Downloads/RVC0813Nvidia/lib/audio.py", line 19, in load_audio
    raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail)

Traceback (most recent call last):
  File "/home/default/.local/lib/python3.10/site-packages/gradio/routes.py", line 321, in run_predict
    output = await app.blocks.process_api(
  File "/home/default/.local/lib/python3.10/site-packages/gradio/blocks.py", line 1007, in process_api
    data = self.postprocess_data(fn_index, result["prediction"], state)
  File "/home/default/.local/lib/python3.10/site-packages/gradio/blocks.py", line 953, in postprocess_data
    prediction_value = block.postprocess(prediction_value)
  File "/home/default/.local/lib/python3.10/site-packages/gradio/components.py", line 2076, in postprocess
   processing_utils.audio_to_file(sample_rate, data, file.name)
  File "/home/default/.local/lib/python3.10/site-packages/gradio/processing_utils.py", line 206, in audio_to_file
   data = convert_to_16_bit_wav(data)
  File "/home/default/.local/lib/python3.10/site-packages/gradio/processing_utils.py", line 219, in convert_to_16_bit_wav
  if data.dtype in [np.float64, np.float32, np.float16]:
AttributeError: 'NoneType' object has no attribute 'dtype'

[candid prairie]

really can rvc able to do that

[radiant flare]

sorry for giant text block

[copper pebble]

dont really know but i suggest to use okada

[viscid moss]

hmmmm

[copper pebble]

https://docs.aihub.wtf/voice-changer-realtime/w-okada-voice-changer-realtime

heres a guide for it

[viscid moss]

Ilaria TTS #1159513888199540817

[karmic locust]

i had tried okada before but i had better results with the realtime capabilities of the RVC1006Nvidia release of voice conversion, which i did have working but something broke

[karmic locust]

i appreciate the thought though

[copper pebble]

theres also this option in our guide https://docs.aihub.wtf/voice-changer-realtime/rvc-voice-changer-realtime

[karmic locust]

yeah that's the one im using that broke :(

[copper pebble]

ohh okok

[karmic locust]

i cannot figure out why it broke or how to fix it, the root of that doesnt contain a venv to cd into

[copper pebble]

seems like theres some trouble with the xformers

[karmic locust]

usually im good about being able to follow dependency hell and making it work, but i'm at the end of my rope here with the xformers issues

[rare goblet]

Ayo? @karmic locust level 2 !!!

[odd shale]

Un día de estos creo que deberías hacer un tutorial de como instalar Mangio Fork en Vast.ai

[karmic locust]

OMG I FIXED IT

there was a random python folder with site packages sitting in the root of my users folder