✨│ai-help

[molten pecan]

Hm I'm a little confused about what you want

[vast forge]

yes or no

[half cove]

you can only use the steps at every checkpoint

[half cove]

yes and please use what is said in the guide

[dawn nebula]

they want to adjust the save frequency so that it matches the recording point on the tensorboard

[molten pecan]

But first, you are aware that you don't need exactly the lowest point of the graph, right?

You won't notice the difference if you pick the closest to it

[placid holly]

picture 1, the tensorboard show 770, picture 2 when saved it show s749

i did get for SVC, but rvc i don't get how to calculate it

[molten pecan]

SVC?

Do you still use svc?

[dawn nebula]

-svc

[azure marsh]

https://cdn.discordapp.com/attachments/1159822429137412177/1194008385113301003/svc.gif?ex=65c13ef6&is=65aec9f6&hm=ce1f57bc316793f641c80f9b359ba9db7d1312af6918b1aed9c8468c74035bac&

[molten pecan]

0cm0evmqpmg0w

😭

[half cove]

it says v2 so im guessing its rvc unless im dumb

[molten pecan]

@placid holly read this

[placid holly]

long back then, i did calculate for SVC and solve it

but RVC, i'm stuck

[molten pecan]

Which version of RVC are you using?

Which fork?

[placid holly]

the Mangio fork, i just try to match the step

nothing any special

sometime they match each other, but i don't know how tf i did

[molten pecan]

Use a save frequency of 10

[vast forge]

how long does training normally take?

[rare goblet]

Ayo? @vast forge level 3 !!!

[half cove]

your using colab so i wouldnt train over like 2h

[vast forge]

so i can just leave it training

[half cove]

you should stay on the tab as it may check if your afk

and make you do a captcha

[vast forge]

its on epoch 50

tensorboard is working now

epoch 100

[severe vapor]

Gen Q abt training an RVC model: I understand that shit isnt as simple as "more epochs = better model", but are there consequences to running more epochs other than wasting your own time? Do models just plateau in quality after a while or could running more epochs somehow make it worse?

[half cove]

too many epochs can be bad due to overtraining

you can just train untill it overtrains and choose the checkpoint right before that

[severe vapor]

How do i know if its overtrained? Just testing and output quality?

[half cove]

-overtrain

[azure marsh]

Hey, @severe vapor!

👇 Here are some resources to help you identify if your model is overtraining

How do I know if my model is overtaining?

All-In-One Guide on how to make a good model

This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[severe vapor]

exactly what i needed, thanks for pointing me in the right direction

[rare goblet]

Ayo? @severe vapor level 1 !!!

[placid holly]

extra just go for 16k/32k

damn, 13 second ++ delay

[teal ridge]

How do I fix that damn issue?

[placid holly]

use gpu, Okada really heavy program

or just use the collab one

[teal ridge]

👌 I'll try it.

[molten pecan]

@tall ocean check my About Me

[tall ocean]

so sorry

[molten pecan]

about your question, you take the .PTH that's closest to before 500 steps

not epochs, but step

those two are different things

[molten pecan]

np

[tall ocean]

Is it listed on the .PTH file the number of steps that it has?

[molten pecan]

yup

check around the end of the training tutorial of your rvc in the page

[tall ocean]

okay, gotcha

something about this graph doesn't look right

[molten pecan]

you think so?

for how long has it been going?

[tall ocean]

we're at 100 epochs

[molten pecan]

ah I see

iirc it looks flat like that when it's starting

[tall ocean]

ah alright, I see

I'm using 8 minutes of training data, no silence

[molten pecan]

which batch size did you pick?

[tall ocean]

I went lower for this one 8

[molten pecan]

sounds good

👍

[half cove]

im guessing thats loss/g/total

[tall ocean]

@half cove told me about the batch sizes haha

nice name btw

for my first attempt at this I used a bad dataset with reverb and silence and it still turned out alright

want to see how far I can push it

[molten pecan]

oh no, reverb?

[tall ocean]

yep...

it was so bad you could even say it was an echo

[molten pecan]

oh you are not training a model with reverb right now, right?

[vast forge]

what does the % mean next to the epoch

[tall ocean]

nope haha, I used rx10 to get rid of that and mastered everything using fl studio

[rare goblet]

Ayo? @tall ocean level 3 !!!

[vast forge]

is it just % done?

[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

[tall ocean]

good question @vast forge no clue

[vast forge]

im making a shotgun willy model

[tall ocean]

If i had to guess it's how complete that specific epoch is

[tall ocean]

nice

I'm making handunit from fnaf

Andy Field voice actor

[vast forge]

I don't really know FNAF who/what is a handunit

[tall ocean]

just listen to the voice https://www.youtube.com/watch?v=lK4PZ9jh-Cg

[vast forge]

OHHH

Fnaf sister location

should i let it go to 1000

[tall ocean]

I had to restart because I realized half of my dataset was missing from the file

[vast forge]

its on 352

i did 1000 instead of 2000

[tall ocean]

I'm following Julia's guide on when to stop training

[vast forge]

how do i stop training

[tall ocean]

So at a certain point the graph lowers and does not lower any more

that's when you stop

https://aihubdocs.github.io/en/rvc-resources/epochs-overtraining--tensorboard/#how-to-install-open

[vast forge]

where on tensorboard do i find that

[tall ocean]

the guide tells you how to do everything

g/total I think is what you are looking for

[proper shale]

Ehhhhh, this is kinda inaccurate.

Overtraining doesn't happen like that, normally once it's been increasing for a while and the other indicators aren't good (g/mel and g/kl), that's when you know it's overtraining. You should look for the latest non overtrained spot (because 99% of the time it's a better model).

But, again, that's advanced stuff

[proper shale]

Oh that's an awful lot of collapses. Were you using Ov2?

[vast forge]

no

i was using original

[proper shale]

What batch size?

[vast forge]

8

[proper shale]

It's most likely a dataset problem... but to certify that I'd try again with rvmpe as your pitch extractor and 4 batch_size

[vast forge]

not a single captcha appeared yet

[vast forge]

i've been doing this for 1 hour and 30 minutes

[proper shale]

You should probably stop & test your model

Collapses are a bit of a pain but they're not that bad if the model keeps improving

[vast forge]

if i stop it will it save

[rare goblet]

Ayo? @vast forge level 4 !!!

[proper shale]

Yeah but you need to run the "Export Model to Drive" step

After stopping the training step

[vast forge]

before or after

[proper shale]

After the training step

And then your training files will be saved and you can continue later

[vast forge]

so end it then export model

[proper shale]

Yup

[vast forge]

ok

export to

wait

what do i change these to

@proper shale

[proper shale]

no need to change anything

just run the step

[vast forge]

ok

i did it

now its in drive?

[proper shale]

now everything should be in ur model's folder in rvcDisconnected

so yeah, its in ur drive

[vast forge]

which one is it

is this good?

[molten pecan]

yes. those are the checkpoints

[vast forge]

do i download any if them>

?

[brittle wing]

anyone know what to use for the tts ai videos? slowly learning about this stuff

[proper shale]

Eh not really, the latest one on your drive is most likely better

[proper shale]

You can use ElevenLabs for TTS

[grim nova]

any sick dev's in here looking to hire !! Huge project !!

[proper shale]

Free, but limited to 10k characters per month

Not related to the voices we got here too

[vast forge]

it might work

it appeared

IT WORKS

[sinful lava]

i need help guys

any clue what to do here

-help

[azure marsh]

Command has expired

[half cove]

What does cmd say

[vast forge]

how do i upload the model to huggingface

[half cove]

https://aihubdocs.github.io/en/essentials/how-to-upload-models-to-hugging-face/

[meager bear]

Hi, I keep getting the access denied error even though i already put my own AI voice in. yes i did put acces to "anyone with the link". used to work before now it doesnt

[empty tusk]

how do i find a collab disconect method?

[rare goblet]

Ayo? @empty tusk level 1 !!!

[molten pecan]

Where exactly?

In Google Colab?

[meager bear]

yeah

if theres a new way im all in

[molten pecan]

Which Colab space exactly?

What's the name?

[meager bear]

Easy GUI (for RVC v2, with crepe) (with improved downloader)

[molten pecan]

https://aihubdocs.github.io/en/ check the Ilaria RVC page

It's up-to-date and fast asf

Check the troubleshooting page too, regarding the model download

[meager bear]

thanks

[rare goblet]

Ayo? @meager bear level 1 !!!

[molten pecan]

yw!

[glad hill]

alguein sabe como hacer para q en el fortnite o otros juegos funcinoe el programa?

ya tengo el cable virtual

[sinful lava]

once i have my models trained in the folder what do i do to make them sing a song

[lyric vine]

Guy help me pls

[molten pecan]

Replace "blob" for "resolve"

;3

[molten pecan]

https://aihubdocs.github.io/en/ check the ai cover page

[molten pecan]

#🔍│help-w-okada

[sinful lava]

do i drag the pth file to audacity

[molten pecan]

No. Please read what I send

[sinful lava]

I did

once you convert vocals

where is the file to mix it

[molten pecan]

Depends

If you are using Colab or local RVC

[sinful lava]

rvc

[rare goblet]

Ayo? @sinful lava level 3 !!!

[empty tusk]

can someone help me with colab rq

[molten pecan]

What's up?

[empty tusk]

how do i find a disconect method?

prevention method*

[sinful lava]

can i pay someone to make one for me

[molten pecan]

https://docs.google.com/document/d/1biKwONvquRJp5nh6jji-AQNrnR5cdyMRciOQMUpNg7c/edit?usp=drivesdk scroll down to "codigo para estar afk"

You'll have to use the translator

[empty tusk]

thank youu!

[molten pecan]

yw! Have fun

[molten pecan]

Yes. #1159289738314919936

[empty tusk]

i was gonna do it locally but my pc is too slow for that haha

[sinful lava]

when it says stop training

where does it say that

[brittle wing]

I came for sound mod but I can't find Turkish models

[raw steeple]

i'm tranning a layne staley model but it sounds robotic, what would you guys do to solve that, more isolated vocals ?

[hot lagoon]

when like makeing a cover what do you guys usually do with the back vocals?

[pine flame]

Is it a bad idea to include some whispering in a dataset? Nevermind removed it

[rare goblet]

Ayo? @pine flame level 2 !!!

[tall ocean]

@pine flame I feel like having a separate model for the same voice but in a whisper would be smart though because whispering doesn't work very well in a real time voice changer

Could someone take a listen to my dataset and tell me if there's something very clearly wrong?

Maybe I can't get good results because I added compression to the vocals or something

[lime finch]

whats a good rvc model thats a girl voice?

[tall ocean]

@lime finch I have 2 of them that don't sound terrible, did you want them?

[forest sluice]

#1175430844685484042 look for any anime girl models

[molten pecan]

@dusty rampart has the best ones. Check his About Me

[dusty rampart]

heya

[molten pecan]

https://ko-fi.com/ilariaowo

This one is pretty good too

[molten pecan]

Go to eep mr admin

[dusty rampart]

2AM

training models

cant

[molten pecan]

Do you sleep

[dusty rampart]

is basically passing out in VR for 4 hours considered sleeping?

[molten pecan]

😭 do you eat?

Or do you do photosynthesis

[dusty rampart]

cant take sun light

i'm doing laser procedures

[molten pecan]

Weird question but ever considered stepping down from ai hub admin for now?

You have a lot of responsibilities in your life

[dusty rampart]

not a role problem

[molten pecan]

Being an admin on two huge servers is another one of em

[dusty rampart]

https://tenor.com/view/gojo-i-will-win-nah-id-win-demonshyne-gif-7267348142363613442

[mortal panther]

I made a really long voice file of my awful singing, i payed 9£ to train them w kits.ai but it keeps crashing

[molten pecan]

@half cove hggghhhgghjhhhj this goes so hard ......

[mortal panther]

How can i like..train a voice on iphoke

[dusty rampart]

💀 Kits.ai

[mortal panther]

Yeah

[molten pecan]

Oh no...

[mortal panther]

I realised how stupid it was

Now

[molten pecan]

I'm so sorry

[mortal panther]

Its okay

[molten pecan]

You know you can train quality models from your phone nowadays, right?

For free too

[mortal panther]

W what

Im bad w technology

Even though i’m 17

Hahahaha

[dusty rampart]

if you have enough patience

it works ig

[molten pecan]

https://aihubdocs.github.io/en/ np. We made a beginner-friendly guide =3

[mortal panther]

Thank u

Lemme checkit out

[molten pecan]

yw have fun!

[lime finch]

yes!

[mortal panther]

I gotta download this right

I have the data

[molten pecan]

This is for desktop though

[mortal panther]

Ohh

[molten pecan]

You'll need good specs for that

[mortal panther]

IPhone 13 Pro Max

Will it do

[molten pecan]

Yeah any phone that has an internet connection can do it

[mortal panther]

Okk

[molten pecan]

It's through a website

It won't consume the resources

Check the RVC Disconnected page

[mortal panther]

I had a cold when i sang so it’s gonna be my voice

But when i have a cold

HAHAH

[lime finch]

I'm just trying to find something realistic

[rare goblet]

Ayo? @lime finch level 1 !!!

[left sky]

I don't know if this is the proper place to ask but I need help with an actual voice model that sounds like a women, not an anime. My voice isn't deep deep but i'm trying to get something that's properly trained and not roboticy aswell.

[lime finch]

im in the same boat lol

[left sky]

Think we have alot of work to get our voices tuned to sound properly.

[hollow dome]

everything is loading forever, is there anything I can do?

[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

[stiff helm]

Does anyone know how I can inference an RVC model on Mac

[forest sluice]

Yes you can

[stiff helm]

How?

[forest sluice]

Do you have python installed?

[stiff helm]

Yes

[forest sluice]

Get version between 3.8 and 3.10, download the https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI repo, open the terminal, use cd to go to the RVC repo, type in this command first:

  python3 -m venv .venv
  source .venv/bin/activate

Then, use pip install -r requirements.txt, and finally, to launch the WebUI, use python3 infer-web.py --pycmd python3

Is your Mac the M1/M2 types?

Those chips are hella performant

[stiff helm]

It's the M2

[stiff helm]

Do you know how I install the repo?

[forest sluice]

There's a "Code" button, download the zip file and extract

Or if you have git, use git clone on your terminal

[stiff helm]

@forest sluice Thank you so much

I'm installing it now

[glad zealot]

does that install all the pip stuff in that folder?

[sand iris]

when you're using a venv everything gets installed there

[glad zealot]

nice

wonder how colab will do with it

[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

[stiff helm]

What's the benefit of using MaglioRVC over the WebUI

[proud elbow]

probably some QoL stuffs and mangio-crepe f0 method (basically crepe with adjustable hop length), though the original RVC has newer latest version

[stiff helm]

Thanks. What does "adjustable hop length" mean?

[proud elbow]

there's mangio-crepe hop length setting in the inference and training UI, if you select mangio-crepe

[stiff helm]

what does hop length mean?

[proud elbow]

adjusting it lower will increase pitch accuracy, but will be slower

[stiff helm]

I see. With the original RVC, that's not possible?

[rare goblet]

Ayo? @stiff helm level 5 !!!

[proud elbow]

it only has crepe method without hop length setting, but anyway rmvpe works better generally (and all recent RVC forks have it)

[stiff helm]

I see, I'll go with rmvpe then, thank you

@proud elbow do you have experience with the RVC webUI? I'm trying to run it for the first time, and having a bit of trouble

[proud elbow]

currently I have RVC on an intel iGPU laptop, and can only run cpu mode
so I have no much idea on your Mac as a different system
what's your issue now tho?

[stiff helm]

Ah that's alright, no worries

I can tell you what's the problem

These are the things I inputted

Both of those files, I put in the directory here

[half cove]

try put the full file path

[stiff helm]

I tried

[proud elbow]

the added_* file is supposed to be in logs\yourmodel

[half cove]

you can use it from path without it being it logs but put in for the dropdown

[stiff helm]

Should I create a "logs\yourmodel"?

I was storing my "experiment_name.pth" file here

[proud elbow]

then put it in logs\experiment_name

I mean the added_*.index file

[half cove]

^

keep pth in weights

index in logs

[stiff helm]

Like this?

[half cove]

yes

[stiff helm]

Great, it looks like this now

I still get the error, I'll print the full error that I'm getting

Thank you guys for the help so far @half cove @proud elbow

2024-02-03 07:28:56 | WARNING | infer.modules.vc.modules | Traceback (most recent call last):
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/infer/modules/vc/modules.py", line 172, in vc_single
self.hubert_model = load_hubert(self.config)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/infer/modules/vc/utils.py", line 23, in load_hubert
models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/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 "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1349, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1283, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/components.py", line 2586, in postprocess
file_path = self.audio_to_temp_file(
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/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'
2024-02-03 07:28:56 | INFO | httpx | HTTP Request: POST http://localhost:7865/api/predict "HTTP/1.1 500 Internal Server Error"
2024-02-03 07:28:56 | INFO | httpx | HTTP Request: POST http://localhost:7865/reset "HTTP/1.1 200 OK"

[half cove]

put the full file path of the audio

"C:\Users\user\Downloads\yes123.wav"

i dont really know if its different on mac

[stiff helm]

This is how paths are written on mac

Unfortunately, I got the same error

[forest sluice]

I forgot to tell you to run dlmodels.sh

[stiff helm]

I can try that. Just "dlmodels.sh" in the commandline?

[forest sluice]

On your terminal, ./tools/dlmodels.sh

It'll also download pretrained model, should you decide to train something

[stiff helm]

Is this a requirement?

It says I must install aria2 to run this (but aria2 fails when I try to install)

[forest sluice]

Yeah

[stiff helm]

Ah, ok

Please set 'WHL_PLATFORM' env var or run this building on supported platform.
supported platform: manylinux_2_17_x86_64, musllinux_1_1_x86_64, manylinux_2_17_aarch64, musllinux_1_1_aarch64, win_amd64, win32

The install fails, seemingly b/c Mac is not supported

"Please set 'WHL_PLATFORM' env var" I'm not sure what this means, but it seems like my best best

[forest sluice]

In your RVC folder, run these commands

curl https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/hubert_base.pt -o ./assets/hubert/hubert_base.pt
curl https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/rmvpe.pt -o ./assets/rmvpe/rmvpe.pt

This will download necessary files for inference

[stiff helm]

Wow, you're actually the greatest

It downloaded both of those files

Thanks sm, I'll try again now

Hm, it looks like the initial hubert issue was solved

[lament ermine]

where i can find tutorials on making models?

[bitter comet]

just wondering how many epocs is just enough for a voice to sound good especially when re creating a rappers vocals?

[half cove]

https://aihubdocs.github.io/en/essentials/how-to-make-an-rvc-voice-model/

[half cove]

do like 500 and check when it overtrains https://aihubdocs.github.io/en/rvc-resources/epochs-overtraining--tensorboard/

[stiff helm]

@forest sluice what is the hubert_base.pt file supposed to look like?

[forest sluice]

Maybe stop the WebUI, delete the hubert file, redownload it, then re-launch the infer-web.py file?

The error seems that the file is corrupted

[stiff helm]

I think you're onto something here, I don't think the hubert_base.pt file is formatted correctly

[rare goblet]

Ayo? @stiff helm level 6 !!!

[stiff helm]

Found. Redirecting to https://cdn-lfs.huggingface.co/repos/ef/97/ef977053f017cde1fc0f89ac7ef0b33172a3d8fb8a840bb24e78b1c0f35f1e72/f54b40fd2802423a5643779c4861af1e9ee9c1564dc9d32f54f20b5ffba7db96?response-content-disposition=attachment%3B+filename*%3DUTF-8''hubert_base.pt%3B+filename%3D"hubert_base.pt"%3B&Expires=1707222354&Policy=eyJTdGF0ZW1lbnQiOlt7IkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTcwNzIyMjM1NH19LCJSZXNvdXJjZSI6Imh0dHBzOi8vY2RuLWxmcy5odWdnaW5nZmFjZS5jby9yZXBvcy9lZi85Ny9lZjk3NzA1M2YwMTdjZGUxZmMwZjg5YWM3ZWYwYjMzMTcyYTNkOGZiOGE4NDBiYjI0ZTc4YjFjMGYzNWYxZTcyL2Y1NGI0MGZkMjgwMjQyM2E1NjQzNzc5YzQ4NjFhZjFlOWVlOWMxNTY0ZGM5ZDMyZjU0ZjIwYjVmZmJhN2RiOTY~cmVzcG9uc2UtY29udGVudC1kaXNwb3NpdGlvbj0qIn1dfQ__&Signature=h~x06z~ocDnAnntt3flKD64MY2sICCCIdZkioTe~9qjASVHpJjt4yQttdbmqMPQs3JAADQvIv7Qfkstji6DUiHvRwDf638G8ohBzQCgMIWwsOMg8W3gOIyFHyEVyIDB~iYYHYNEYyTljFliM4MIrErBnZWvVcEtt5yj3KXGgfme~iBuXYPLsbHGJHUUy8oh68PN9ZIEQdId0OhTFj78yHjPmdGnKHfQ1uioTKpdU4rCOjLsr2F~Pd8-KtALGXkfUa~7D1xgUdOocAmVuAicIoK5HTIe-MbV5DP4hGOEdRo1J-yt5i7hSECeTjpK0vA0NCtaszDNzuH2EqptABqpWaQ__&Key-Pair-Id=KVTP0A1DKRTAX

This is all thats in it

[bitter comet]

so what i kind of am confused about mainly the overtraining what will i need to do when it overtrains do i need to stop it and then will it be good or do i need to download more samples like Acapellas and add it and train it on that as well

[stiff helm]

@forest sluice I fixed that error!

I am on to a new one though:

File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/infer/modules/vc/pipeline.py", line 154, in get_f0
f0 = self.model_rmvpe.infer_from_audio(x, thred=0.03)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/infer/lib/rmvpe.py", line 599, in infer_from_audio
mel = self.mel_extractor(
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/infer/lib/rmvpe.py", line 472, in forward
fft = torch.stft(

File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/torch/functional.py", line 660, in stft
return _VF.stft(input, n_fft, hop_length, win_length, window, # type: ignore[attr-defined]
NotImplementedError: The operator 'aten::_fft_r2c' is not currently implemented for the MPS device. If you want this op to be added in priority during the prototype phase of this feature, please comment on https://github.com/pytorch/pytorch/issues/77764. As a temporary fix, you can set the environment variable PYTORCH_ENABLE_MPS_FALLBACK=1 to use the CPU as a fallback for this op. WARNING: this will be slower than running natively on MPS.

Traceback (most recent call last):
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1349, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/blocks.py", line 1283, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/site-packages/gradio/components.py", line 2586, in postprocess
file_path = self.audio_to_temp_file(

File "/Users/matthewpena/Documents/Archived/Code/tools2/Retrieval-based-Voice-Conversion-WebUI/.venv/lib/python3.10/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'
2024-02-03 08:15:57 | INFO | httpx | HTTP Request: POST http://localhost:7865/api/predict "HTTP/1.1 500 Internal Server Error"
2024-02-03 08:15:57 | INFO | httpx | HTTP Request: POST http://localhost:7865/reset "HTTP/1.1 200 OK"

[half cove]

id train more than it overtrains then on weighs choose the checkpoint to the closest one to right before it overtrains make sure you have save small final model option selected then every time it saves it will create a model

[forest sluice]

you have to set the environment variable like what the error says: run export PYTORCH_ENABLE_MPS_FALLBACK=1 before launching the WebUI

[stiff helm]

Thanks so much Ouch

I'll do that

[bitter comet]

ah okay also noticed they told me about a time limit and switching accounts if i upgrade to pro for example will i be able to go for longer period of times?

[half cove]

if your using colab try limit training to 2h or under

or it might reach the gpu limit and kick you

if you upgrade to pro it will give you compute units

and will kick you after those are done

[bitter comet]

and for how long can i go then?

with the units

[half cove]

i dont really know tbh

[bitter comet]

so basically after like 2 hours i just press the square to stop it and continue after 2 hours again?

or is there a 24 hour cooldown?

[half cove]

after the 2 hours you shouldnt train it more or it might kick you

you can use alts

[bitter comet]

alright

[prisma hearth]

Is it possible to create an AI voice model with a mobile phone?

[proud elbow]

https://discord.com/channels/1159260121998827560/1160926856548200531
you can run the colab through a phone

[prisma hearth]

How can I solve the problem of mobile phones being interrupted during production? I tried it a few times, but I got frustrated and gave up.

[proud elbow]

wdym being interrupted, elaborate?

[prisma hearth]

?

[prisma hearth]

휴대폰은 이방법을 할수 없어서 중간에 중단됩니다.
Cell phones cannot do this, so it stops midway.

[proud elbow]

not really sure about the devtools in phone browser, probably try google some guide on it

[half cove]

just stay on the page and check up on it

[prisma hearth]

Of course, it's impossible without a computer.

[half cove]

that doesnt work much anyway as now google does captchas

[prisma hearth]

No matter how many times I check and click...

[half cove]

also use https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnected/ so the images work correctly

[half cove]

?

[prisma hearth]

I tried all sorts of things, but gave up due to an error with my phone disconnecting in the middle.

[half cove]

1. did you leave your phone on without checking it every few mins
2. how long did you train

[prisma hearth]

The mobile phone was moved by hand every five minutes. I tested it at around 300, but an error occurred in less than 30 minutes...

[proud elbow]

your phone network was unstable, probably?

[prisma hearth]

I already saw it.

[half cove]

if the network was unstable it should still run but it will disconnect if your gone for too long

[proud elbow]

the usage time is usually up to 4 hours, even on free account

[prisma hearth]

The maximum speed of Wi-Fi is over 500MB..

[half cove]

what was the message it gave you when it disconnected you

[prisma hearth]

An error occurred.

[proud elbow]

the network bandwidth isn't always relevant to occasional ping spikes tho

[half cove]

an error occurred?

now thats new

i havent seen that before

like while training?

[prisma hearth]

Anyway, this method doesn't work with a cell phone...T.T

아무튼 이 방법이 휴대폰으로 안된다...T.T

[half cove]

it should work on cell phones

[prisma hearth]

I understand what everyone is saying. As I've said many times, it's annoying because of errors that cause it to stop in the middle.

[proud elbow]

android phone tho?

[half cove]

it should work on basically anything that has internet

[prisma hearth]

I just think it would be easier for me to give up. thank you for telling me.

[proud elbow]

what kind of error details?

[prisma hearth]

yes. Samsung Galaxy S10e

[glad zealot]

i got that error a few times on pc.. when something runs that somehow uses more resources than its supposed to.. like more ram thatn its supposed to

[half cove]

would a lower batch fix it?

[glad zealot]

maybe?

[cobalt saddle]

what is this?

[worldly citrus]

oh

it's like dynamics of the voice

for me best option is default

[cobalt saddle]

to ile ustawić?

[rare goblet]

Ayo? @cobalt saddle level 3 !!!

[cobalt saddle]

@worldly citrus

[worldly citrus]

128

[rare goblet]

Ayo? @worldly citrus level 2 !!!

[worldly citrus]

to wartość domyślna

[molten pecan]

Uhm are you using rvc gui?

[cobalt saddle]

yes

[molten pecan]

Oh no that's really outdated

[cobalt saddle]

where can i get latest?

[cobalt saddle]

found it on yt im new to this shi

[molten pecan]

Do you have an NVIDIA GPU?

[cobalt saddle]

yes

[molten pecan]

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006AMD_Intel.7z

Click download

https://aihubdocs.github.io/en/

Check the "mainline rvc" page for a usage guide

[cobalt saddle]

okay , thank you ;DS

[molten pecan]

yw!

[azure patio]

is there any link for a working rvc?

[half cove]

-links

[azure marsh]

Suggestions for @azure patio

All Working Local Forks 🖥️

• 🍏 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 "❔┋guides" channel. If you can't find your answer, feel free to ask for help in "help-rvc". Credits to Faze Masta and Antasma for compiling these links.

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

[azure patio]

thanksss

[rare goblet]

Ayo? @azure patio level 2 !!!

[sand iris]

that's the amd download !!!!

On top of being unusable it's the wrong one

[molten pecan]

OH FUCK

😭

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006Nvidia.7z @cobalt saddle my bad, this was the right link

Sorry

I misclicked

[molten pecan]

Thank you GDR

[sand iris]

https://tenor.com/view/moistcritikal-penguinz0-charlie-misinput-it-was-a-misinput-gif-23090919

[molten pecan]

[molten pecan]

Why do they still have it if it's unusable?

[sand iris]

only for real time

I really don't know, they removed it randomly for no reason

[molten pecan]

Weird

[low python]

Guys, when using AICoverGen-WebUI, modded by Hina is there a way to have the back-up vocals follow your voice model too. I find that the UVR and conversion work absolutely wonderful yet f.s.r. the back-up vocals always stay like the original. Only the main converts. This can be an issue because while the UVR is good it doesn't always distinguish main vocals and back-up vocals correctly

Is there a way to have it convert both main and back up at the same time?

[proper shale]

Nope, that's just how it's programmed to be

[low python]

wonderful

[proper shale]

Just better to filter the stuff out yourself

[low python]

yeah been doing that. It kinda works it's just. eh.

It's hit or miss really

On one hand I have been able to make better covers than ever, on the other hand I struggle.

Of course not all songs are coverable

[proper shale]

yeah

[vast forge]

what happens if i stop the training at epoch 50

would it just sound weird

[frank roost]

Anyone know any good rcv v2 voice trainers? easy doesnt work anymore and im trying to clone my voice

[proper shale]

eh not really

i mean, would probably sound ok but it could improve yk

[proper shale]

RVC disconnected, check https://aihubdocs.github.io/en

[frank roost]

thanks right before u sent that i litterly just ran the cell lol

[rare goblet]

Ayo? @frank roost level 1 !!!

[proper shale]

oh damn

[vast forge]

im trying a ai voice of one of my friends

i asked them before making them

a ai

[vast forge]

is it possible to make the ai voice a voice changer?

[rare goblet]

Ayo? @vast forge level 5 !!!

[proper shale]

u can use W-Okada to change ur voice w the model yeah, you do need a good GPU or u could use colab

[stiff helm]

@forest sluice

Ok, so I let it sit for a while

I never got an error, but it just never finished

We're really close, but there's got to be some reason it hasn't finished yet

I only have 3 min 47 sec of audio inputted

[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

[vast forge]

can you send the link

[proper shale]

https://colab.research.google.com/github/hinabl/voice-changer-colab/blob/master/Hina_Modified_Realtime_Voice_Changer_on_Colab.ipynb

also check https://rentry.co/VoiceChangerGuide

[stiff helm]

@forest sluice are you sure that rmvpe.pt is correct here?

It seems like that's what its loading

[stiff helm]

^

[dawn nebula]

still loading?

[stiff helm]

Yeah

I tried re-doing it, same problem

[dawn nebula]

are u using linux

or windows

[stiff helm]

Mac

[dawn nebula]

what gpu do u have

[stiff helm]

Tesla T4

[dawn nebula]

so u running colab

[stiff helm]

No, I downloaded RVC

There are options to download RVC on Mac

[dawn nebula]

yeah ik that

coz tesla t4 is the free colab gpu sry i misunderstood💀

[stiff helm]

Oh np

[dawn nebula]

did u install via prezip

or git clone

[stiff helm]

git clone

I did git clone to donwload the repo, I'm not sure what prezip is

[dawn nebula]

so u have installed everything, including pretrains and huberts?

[dawn nebula]

like a .7z that can be used after u extract it

[stiff helm]

I did git clone, and then I manually installed hubert_base.py, and rvmpe.pt

[dawn nebula]

fair enough

[stiff helm]

I'm sorry that it's all kind of scattered, but I was following @forest sluice 's advice

[dawn nebula]

you need training or inference

or both

[dawn nebula]

yeah i saw that

./assets/hubert/hubert_base.pt

./assets/pretrained

./assets/uvr5_weights

Additional downloads are required if you want to test the v2 version of the model.

./assets/pretrained_v2

If you want to test the v2 version model (the v2 version model has changed the input from the 256 dimensional feature of 9-layer Hubert+final_proj to the 768 dimensional feature of 12-layer Hubert, and has added 3 period discriminators), you will need to download additional features

./assets/pretrained_v2

#If you are using Windows, you may also need these two files, skip if FFmpeg and FFprobe are installed
ffmpeg.exe

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffmpeg.exe

ffprobe.exe

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffprobe.exe

If you want to use the latest SOTA RMVPE vocal pitch extraction algorithm, you need to download the RMVPE weights and place them in the RVC root directory

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/rmvpe.pt

For AMD/Intel graphics cards users you need download:

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/rmvpe.onnx

check if everything has installed

[stiff helm]

Thank you, I'll check now

./assets/pretrained

./assets/uvr5_weights

./assets/pretrained_v2

I have these folders, but there's nothing in them except a .gitignore

[dawn nebula]

download it from huggingface

[stiff helm]

Download what? The contents of the folders?

[dawn nebula]

yeah

[stiff helm]

Ok, I'll look at that

Also,

For AMD/Intel graphics cards users you need download:

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/rmvpe.onnx

Do you know where I should store this?

[proper shale]

rvmpe.pt will work fine most likely, u dont need onnx

[stiff helm]

It's not loading in it seems like

[stiff helm]

Do you have a link to the huggingface

[proper shale]

i mean, if its in the right place, it most likely is being loaded

so idk

[sand iris]

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/tree/main/tools

better yet

create a .bat file with the following :

runtime\python.exe tools\download_models.py
pause

Waiiiit

[half cove]

they are using mac

i think

[sand iris]

no

we're git cloning for some random reason

oh nvm

it is mac

welp, almost the same thing then

python tools/download_models.py (with working dir being RVC)

I couldn't find anything about the way this has been installed, but if you've messed with venv, make sure that's active

[stiff helm]

Got it, installing now

Thanks

[vast forge]

how do i do tts

for free

[stiff helm]

I'm downloading what @sand iris recommended rn, but I think I'm going to need to download this too

Apparently Macbook uses AMD

[sand iris]

I don't think the main RVC project even does onnx

[sand iris]

That would only be for the older desktop macs

else it's always intel or apple hardware

[stiff helm]

Let me check

I have: MacBook Air (M2, 2022)

[sand iris]

That's apple hardware

[vast forge]

how doi do tts

[stiff helm]

Oh, ok

I'll lyk when the command you sent is done with all the downloads

@sand iris ok, finished downloading, ill try again

How long should I expect it to take?

@sand iris I've been running it for 2 min so far, and it seems like the same weird issue

[errant palm]

how to save ai voice??

[rigid mango]

Dumbass don't know the basics

[errant palm]

You're itching for a beating

Nostalgia aint u 4'11

[rigid mango]

I bet your black

[errant palm]

And what if i am

Its black history month

You should get on your knees and suck my dick

[rigid mango]

I'll throw at you kfc chicken

[errant palm]

Ill roundhouse kick your fatass

Nostalgia how old are you arent you 43

[rigid mango]

Bro you sound like a 3'11 kid trying to be Tough 😭

[rare goblet]

Ayo? @rigid mango level 4 !!!

[half cove]

tf going on here

[errant palm]

Nun yo biz damn

[rare goblet]

Ayo? @errant palm level 2 !!!

[errant palm]

Look the other way

[half cove]

it kinda is tho

[errant palm]

Ok "poopmaster" the cesspool calls for you

[rigid mango]

Blue shit tryna be nice

[half cove]

😭

[errant palm]

bro shut the fuck up

[rigid mango]

I'll break your jaw you cum seeker

[half cove]

ok no more

😭

[errant palm]

delete nostalgias message

[half cove]

next person gets mute

[rigid mango]

Ok sorry

[errant palm]

❤️❤️❤️ damn girl u cute ASF! Whats yo numbah! 📲❓

[rigid mango]

Bro show him respect

[errant palm]

I am

[rigid mango]

Blue shit not trying to get your numbah😭

[half cove]

if your gunna meme do it in #✦│chat so people who ask questions can be seen

[rigid mango]

Ok bet

[errant palm]

?? Wdym by meme bro this freak ass nostalgia hopped on my dick first

[modern oxide]

when people say stop you stop

[modern oxide]

https://cdn.discordapp.com/emojis/1166378587448938679.gif?size=48&name=cutedog&quality=lossless

[cosmic tide]

I want to make rvc but when i click "Process data" is not showing anything in "output information" plz help me

This is my first time making rvc

[fluid matrix]

Are you using local or colab?

I assume you are using mainline rvc?

[cosmic tide]

Idk, im checking

[rare goblet]

Ayo? @cosmic tide level 1 !!!

[fluid matrix]

Sometimes clicking on that cmd works lol

Then it resumes

[cosmic tide]

https://www.huggingface.co/lj1995/VoiceConversionWebUI/tree/main

Im using this thing

[fluid matrix]

So local

But this is old

-rvc

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[fluid matrix]

Try the 2nd one

[cosmic tide]

I need to use this thing?

[fluid matrix]

Mainline rvc

[cosmic tide]

Ok

[fluid matrix]

Your link was a old version from may idk how you got that

Probably a random yt tutorial

[cosmic tide]

Yes

So i need to delete the thing that im using?

[fluid matrix]

I would delete that version and install the newest version of mainline following that guide yes

[cosmic tide]

Ok

So just minute

This link right?

https://rentry.co/TrainingVoiceModels

[fluid matrix]

Yes

[cosmic tide]

Ok

Maybe it will take some time because i have old leptop

[fluid matrix]

https://aihubdocs.github.io/en/

This might help too

[fluid matrix]

Do you know your laptop specs? Gpu, ram, CPU?

[cosmic tide]

I dont remember

[half cove]

check task manager then performance

[cosmic tide]

And i dont remember how to check it

Windows r and...

[fluid matrix]

If it's old and you don't even know the specs then it's probably not good enough for local training tbh

[fluid matrix]

No just open task manager

[cosmic tide]

Ok

Im using windows 10

[half cove]

your gpu prob cant run it if its a old laptop that you dont know about you can try colab

[fluid matrix]

Yeah

[fluid matrix]

Check this link for a colab guide

[half cove]

https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnected/

[cosmic tide]

Which thing i need to click?

Ok i found it

[half cove]

brother did not read what we said

at all

[fluid matrix]

Yeah no local training for you

Like we said, please check that link

[cosmic tide]

Ok

[fluid matrix]

Read the new guides and try it out first

[cosmic tide]

Ok

[fluid matrix]

Check for cloud training/colab

[cosmic tide]

That thing?

[rare goblet]

Ayo? @cosmic tide level 2 !!!

[fluid matrix]

Bro please read

[cosmic tide]

Ok

[fluid matrix]

Are you even trying to train a model?

[cosmic tide]

Yah

[fluid matrix]

Yeah then that's the right one

[cosmic tide]

Im doing what the guy doing in the tutorial

[fluid matrix]

The guide right

[half cove]

go on the rvc disconnected tab

[fluid matrix]

Not the video

[cosmic tide]

What?

[fluid matrix]

😊

What's the issue?

[cosmic tide]

Open the tensorboard?

[fluid matrix]

The guide should tell you what to do step by step just follow that

Bro this isn't a 1min process

You have to read

And follow instructions

[cosmic tide]

Ok

[fluid matrix]

It tells you everything very clearly

[cosmic tide]

Ok

[fluid matrix]

Have you even made a dataset yet

[cosmic tide]

I just opened the tensorboard

[fluid matrix]

Bro

What

[cosmic tide]

So i need to download

[half cove]

that doesnt answer the question

no

[fluid matrix]

Please

Read

[half cove]

do you have your dataset

the audio that your going to use to train

[cosmic tide]

Yes

I have a aodio

Audio

[sand iris]

Sorry I'm back. What's going on ?

[cosmic tide]

So i clicked on a key but its not doing something, maybe its because of my leptop

[fluid matrix]

It's because you are very impatient and won't read

Please read the guide

Step by step

You haven't even made a dataset, how do you expect to make a model

What of?

[stiff helm]

Thanks, np

[cosmic tide]

I want to make rvc of sadie adler

[stiff helm]

Basically, the same thing still

Everything downloaded that you sent

[sand iris]

What was happening before ?

[stiff helm]

But same error

Let me show you

It gets stuck on this (seemingly forever)

[sand iris]

https://cdn.discordapp.com/emojis/726807504876601395.webp?size=48&name=hmmmm~2&quality=lossless

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/1269

Seems like you aren't the first

[molten pecan]

@stiff helm

GDR remember to ping people so they see the msg

[sand iris]

no fixes there unfortunately

[molten pecan]

_<

[sand iris]

I tend not to when I reply fast

• I don't have a concrete answer yet

[stiff helm]

Ah, well, thanks

[cosmic tide]

It tell me to open that thing that i downloaded open the TensorBoard, its tell me to press a key, im doing that, this thing closing itself and there is nothing

[fluid matrix]

No it doesnt

Take a deep breath

And read carefully

https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnected/

[cosmic tide]

You send me to this

Right?

[sand iris]

Really not sure what the hell is going on with it

[stiff helm]

Maybe its an issue with rmvpe.pt?

[sand iris]

how big is the file ?

[cosmic tide]

Ok, im an idiot, you are smart, king

[half cove]

he didnt....

[cosmic tide]

Yah i know

I thought he send me to that thing

But im stupid

[cosmic tide]

.

Sorry for destroying your brain cells

[stiff helm]

The audio file?

[sand iris]

rmvpe.pt

[stiff helm]

181.2 MB, and this is what the folder looks like @sand iris

[sand iris]

looks good on that end

[stiff helm]

The audio file is 10.1 MB, if that matters

[sand iris]

idk what could be happening

[stiff helm]

I could try this? @sand iris

[fluid matrix]

https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnected/

i sent you this

3 times

just read that 😭

why do you keep opening random stuff

[cosmic tide]

.

[rare goblet]

Ayo? @cosmic tide level 3 !!!

[fluid matrix]

what

bro its not my fault you dont listen

we are trying to help you

[cosmic tide]

Yah i know

[fluid matrix]

alright then

hope it works out

[cosmic tide]

.

[fluid matrix]

nah all good

[cosmic tide]

Me too

[fluid matrix]

you can do it

[cosmic tide]

Ok

I need to download that?

[stiff helm]

yes, run all of those things

[cosmic tide]

All?

[half cove]

please do read the guide

[half cove]

yes

you can just

close the dropdown

and

[stiff helm]

yeah the guide will show u

[cosmic tide]

Ok

[stiff helm]

i just did this for the first time yesterday too

[half cove]

it will give a option to do em all at once

[stiff helm]

@half cove do you have any idea about this issue

[rare goblet]

Ayo? @stiff helm level 7 !!!

[half cove]

are you the guy whos on mac?

[stiff helm]

yep

[half cove]

i dont know anything about using rvc on mac xd

[stiff helm]

The same issue happened to someone on windows too apparently

[cosmic tide]

What is the name of the folder?

[half cove]

what do you mean by this?

the rvcDisconnected folder?

[cosmic tide]

This is the name of the folder?

That i need to put the audio

[sand iris]

It won't be able to load that

[half cove]

have you read step one?

😭

[stiff helm]

wdym

[sand iris]

I'm not sure they have the code hooked up to load onnx models like that

Yeah no it'll only do it on windows

[stiff helm]

Ah ok, thanks. Where did you see that?

[fluid matrix]

no

[sand iris]

Looked at the RVC code

Tried seeing where it could be stuck with rmvpe earlier but that leads nowhere

[half cove]

[fluid matrix]

he basically gave up and told me to go away 5min ago 💀

[stiff helm]

Ah, well, thanks. I'll keep trying I guess

[fluid matrix]

but i have faith

[cosmic tide]

Do i need to click them?

[proper shale]

yes, preprocessing and feature extraction are essential for training

and saving it to drive allows u to retrain later.

[cosmic tide]

When i click the preprocessing its doing a error

[proper shale]

you didn't run Set Training variables

[cosmic tide]

Oh

Thanks

Its working

[cerulean dew]

is there any site i can use that has the ability to split vocals within a duet so that I can get two AI to do a duet?

[cosmic tide]

Now thats happend

[cerulean dew]

how did u get ur file to a wav?
if u did it with anything other than audacity it fails

[cosmic tide]

https://vocalremover.org/

[half cove]

that isnt it

[cerulean dew]

it can split voices? i thought it just split instrumentals

[cosmic tide]

Yes

[half cove]

you could try uvr bve or 5_hp

[cosmic tide]

What i need to do here?

[fluid matrix]

read

the

guide

[cosmic tide]

I

Did

[half cove]

did you put Set Training Variable name as the same as your dataset folder

[cosmic tide]

I need to do with .zip?

[half cove]

nope

[fluid matrix]

you skipped one step

[half cove]

without the .zip

[fluid matrix]

you didnt feature extract yet

[half cove]

that problem does happen to people who do it correctly btw

[fluid matrix]

it even says that here

[half cove]

either its the zip being formatted incorrectly or it not being named properly

[dawn nebula]

yeah defs dataset issue

[fluid matrix]

i know but in this case he did actually

[half cove]

welp

see

[fluid matrix]

im pretty sure he skipped something at the very beginning

[half cove]

the thing is index training is goofy on colab

it gives that error for many reasons

like the ones i said before

[cosmic tide]

I chqnged the name

And no

Its not working

This is alright?

[rare goblet]

Ayo? @cosmic tide level 4 !!!

[half cove]

is your dataset folder named sadie?

[cosmic tide]

Yes

[half cove]

wait

[half cove]

give me this error

[cerulean dew]

experiment name literally does not matter

its just what u want ur model to be called

i never match it and my models work fine

[half cove]

on your preprocessing and feature extraction cell

not this

[cerulean dew]

this looks exactly like the error i got when i converted my mp3s to wavs with cloudconvert instead of audacity

[rare goblet]

Ayo? @cerulean dew level 6 !!!

[cosmic tide]

And what i need to do?

[cerulean dew]

put the files ur using in ur zip folder through audacity and export them from there before trying again

[stiff helm]

@sand iris im gonna send you a dm

[cerulean dew]

ur also probably gonna need a new zip folder with the corrected files

[cosmic tide]

Ok

So put the aidio in audacity, after editing that and download again the audio in need to zip it and put in there?

[cosmic tide]

Am i right?

[stiff helm]

2024-02-03 16:28:42 | INFO | configs.config | No supported Nvidia GPU found
overwrite preprocess and configs.json
2024-02-03 16:28:42 | INFO | configs.config | Use mps instead
is_half:False, device:mps
2024-02-03 16:28:43 | INFO | main | Use Language: en_US
Running on local URL: http://0.0.0.0:7865

BTW, when I first run it, I get this message about no supported GPU

[cerulean dew]

yes

[cosmic tide]

Ok

[cerulean dew]

yeah thats exactly the message i used to get and its complete bullshit and incorrect and it took me 3 days to find out what the issue was when this happened to me

[cosmic tide]

So how do i delete the audio from the google drive?

[cerulean dew]

it doesnt matter bc u will just have to change the folder name u put into the training thing

[stiff helm]

What was the issue? I am running on Mac (10-core gpu)

[cosmic tide]

Ok

[cerulean dew]

not the GPU lol
it was bc the files werent being processed correctly when a different converter was used on the files (cloud convert)

[sand iris]

That's the usual message on mac ! Use mps instead is a good sign, and what you'd want to see

The "no supported GPU found" Message on the UI is also normal, only inference is supported with macbook's GPUs

Training isn't

[stiff helm]

Huh, so that looks normal? Weird

[stiff helm]

What format was your audio in?

[cerulean dew]

wav

[stiff helm]

Mine too

I actually also used cloud convert

I could try the original mp3

[cerulean dew]

nah just run it through audacity it works and its quick

[stiff helm]

mp3 wont work?

[cerulean dew]

ive never tried it but i do know wav and flac r the preferred file types

[proper shale]

mp3 prolly works cuz ffmpeg, but just do wav

[stiff helm]

Ah, I tried mp3, I get the same (seemingly) infinite loading thing

[serene citrus]

can anyone help me? I am getting this error every time i am trying to convert my vocals to a voice model

[molten pecan]

is that huggingface?

[serene citrus]

yes

https://huggingface.co/spaces/TheStinger/Ilaria_RVC

[molten pecan]

you are using an outdated version

[serene citrus]

this

[molten pecan]

https://aihubdocs.github.io/en

check the ilaria rvc page there

[serene citrus]

I signed up and all, but where is the ilaria section? I can only find "making a model" but i already have one, that i want to convert to some vocals

[molten pecan]

follow this path

RVC > Cloud > Inference > Ilaria RVC

[serene citrus]

it doesnt say "RVC" anywhere

[molten pecan]

oh I meant in the aihubdocs page

[molten pecan]

this one

[serene citrus]

thanks, im trying to upload the voice model now

lets see if it works 🤞

[molten pecan]

have fun!

[serene citrus]

❤️ ❤️ ❤️

its fire

[dawn nebula]

yes, but sorry i have to delete this file because we cannot share songs with instrumentals

copyright strikes hard

[serene citrus]

its alr, my bad

[dawn nebula]

all g bro

[modern forum]

how to add new voice on a voice changer

[rare thorn]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface 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
• GPT-SoVITS (demo), by litagin Huggingface Spaces

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

[static gyro]

Quick question that I forgot the answer to. Doesn't rvC have an inference in hugging face?

[half cove]

elaborate

what do you mean

[static gyro]

Can you infur voices using a hugging face space?

[half cove]

yes

[static gyro]

Where is that?

[half cove]

https://huggingface.co/spaces/TheStinger/Ilaria_RVC

[molten pecan]

Poop the hf space is outdated and slow

https://aihubdocs.github.io/en/ @static gyro check the Ilaria RVC page

[half cove]

they did ask about it

[static gyro]

i need a browser place when my gpu is being used up be rendering sfm frames

[molten pecan]

Yea but it's obsolete at this point

[deep drift]

anyone can help me where i can find Greek models? Its really hard and havent found a way to create any on Linux

[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

[eager condor]

hey can i get help downloading the right real time voice changer for my pc please

[brave garnet]

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs
Version 17b cuda

Download for AMD GPUs
Version 17b directml

Download for Intel GPUs
Version 17b directml

Download for Mac
Version 17b Mac
⠀

Made with love by Ilaria 💖

[proper shale]

@eager condor this should help :)

[fickle delta]

Why do so many models seem to slur words, like they have a lisp?

[spare shale]

Some words (Original post) are not Very Good or Model not for this Language?

This Voice is Russian Speaking but it works in other languages

[spare shale]

But where he sings about cola He's a little bad

[half cove]

copyright?

[spare shale]

I think not here Because 1 This is a different voice acting
2 Is this a Pizza Advert?

But I'm not sure about that

[half cove]

adverts usually use premade songs

[spare shale]

But if necessary I can delete

[half cove]

its not the voice its the instrumentals that get us in trouble

[spare shale]

Yes, I understand

[frosty kayak]

ok does this mean to throw in the towel and start from scratch....im at 150 epochs right now after 6-7 hrs

[fickle delta]

How would I go about making actual quality AI voice covers?
It seems to be mostly a gimmick for me at the moment due to the various weird artifacting, but I've seen some really genuinely awesome covers before and I'd like to sink a bit more time into making the outputs sound good. Any tips?

[rare goblet]

Ayo? @fickle delta level 1 !!!

[spare shale]

Sorry, I don't understand this

[spare shale]

-rvc

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[spare shale]

https://www.weights.gg/

[spare shale]

Here you can create quickly

-local

[azure marsh]

All Working Local Forks 🖥️

• 🍏 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 "❔┋guides" channel. If you can't find your answer, feel free to ask for help in "help-rvc". Credits to Faze Masta and Antasma for compiling these links.

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

[spare shale]

-colab

[azure marsh]

☁️ Google Colabs

• 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
• RVC Disconnected, by Kit Lemonfoot Google Colab
• easyGUI, by rejects Google Colab

🤗 Hugginface 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
• GPT-SoVITS (demo), by litagin Huggingface Spaces

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

[spare shale]

Colab or Locally if you only want to change the vocals (it is not possible to save only the vocals via the website)

[frosty kayak]

its just showing collapses...there are three...i was the guy with the excessive dataset...but i want to keep on going, just not sure if i will get a crappy model if so because of the crashes

[fickle delta]

damn cuh spelling mistake on the first page of the wiki

ouch-

[spare shale]

@brittle wing

[brittle wing]

I don't handle the website 😭

[spare shale]

Forgive me but I don’t understand this table

[spare shale]

T-T