✨│ai-help

[brittle wing]

nvm its not working

can anyone help?

[brittle wing]

can anyone help me with delay issue? idk if its the matter of microphone itself or settings

[teal dome]

when i try using fcpe in applio, i now always get a “FCPEF0Predictor” error saying “unexpected argument ‘sample_rate’” which wasn’t happening before. is something wrong with applio or is it me? the other algos work fine

[rare goblet]

Ayo? @teal dome level 7 !!!

[brittle wing]

-sovits

me irl

[low shard]

are you sure you are using https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp=sharing and not another aicovergennoui?

[pastel oak]

Depends on your gpu

type in #🔍│help-w-okada your gpu name and send a pic of your settings

[brittle wing]

alr

[rare goblet]

Ayo? @brittle wing level 3 !!!

[deep oasis]

can someone teach me how to generate an AI kpop voice? i’ve been trying a lot but I failed 🥹

[low shard]

you want to train an RVC Voice model? check our docs https://docs.aihub.wtf/essentials/how-to-make-ai-cover/

If you got a good pc use Mainline or Applio, else use the Google Colabs, or for a better free cloud experience but harder read the Mainline Kaggle Guide

[void dome]

guy i need help

voiche chger dont work

he makes sounds but not what i say

[pastel oak]

whats your gpu name cuz you didnt select one

[proven hill]

cpu selected

[brittle wing]

doesnt let me post a picture but ineed help

getting this error

content/voice-changer/server

ModuleNotFoundError Traceback (most recent call last)
<ipython-input-6-c4b2eef38e98> in <cell line: 24>()
22 get_ipython().run_line_magic('cd', '/content/voice-changer/server')
23
---> 24 from pyngrok import conf, ngrok
25 MyConfig = conf.PyngrokConfig()
26 MyConfig.auth_token = Token

ModuleNotFoundError: No module named 'pyngrok'

---

NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.

To view examples of installing some common dependencies, click the
"Open Examples" button below.

[brittle wing]

im using the old one ig, it worked like months before I took a break from youtube so I havent been using it now that Im back to making videos it wont let me

[low shard]

send google colab link

[brittle wing]

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

[low shard]

its better u use the newer one that is fixed

[low shard]

dont use that

-rt

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[low shard]

use https://docs.google.com/document/d/e/2PACX-1vTIceEcBfS6Zqolv_QEysrFfI_EJikPxozWptP_EjkpLVl-l-gdo-ijBonQMTviAHEYm5emmd9k9TdC/pub instead

[unborn sigil]

should I keep training or stop?

[glad zealot]

looks good, you can save that model but still keep training if it goes down more

but it does already look good tho

[unborn sigil]

to keep training it, I would set all the setting in applio again, then jsut press begin training right?

no need to do other buttons

[proven hill]

yes

[glad zealot]

yup, just make sure its the original logs file, name, settings

[unborn sigil]

and like what step in the graph should I use to inference with?

or weight

i'm thinking right around here

[glad zealot]

try the latest one

[unborn sigil]

oh ok

[glad zealot]

since thats seems to be your lowest point

[unborn sigil]

the guide says not to do that, but that sounds like a better idea

oh! i read it wrong. Thanks

[brittle wing]

do i upload the python file into model or index>

[proven hill]

what

[glad zealot]

???

[brittle wing]

i cant send it here

[proven hill]

im confused ngl

[unborn sigil]

wow! all those hours of training worked out. I made a model from all my past vocal takes and it sounds decent. Thanks again y'all

[glad zealot]

headpat~2

[proven hill]

[brittle wing]

whys the voice uploading keep changing from 5% to 17% to 29% back down to 16%

[rare goblet]

Ayo? @brittle wing level 1 !!!

[dry bane]

does anyone know how to separate overlapping vocals from the acapella of one song? (say someone sings another lyric while the other person sings)

[pastel oak]

Those are hard to separate, u can try the Karaoke model on mvsep.com

Theres some others to try, someone else might know more

Same applies on UVR

[dry bane]

got this, thank you!

[brittle wing]

i dont understand how to use the voice on discord or something

[odd shale]

You can try using UVR BVE on MVSEP or BV2 V2/Mel Karaoke on X Minus (tho this one is paid)

[pastel oak]

download Virtual Cable, explained on the guide step 6

https://rentry.co/VoiceChangerGuide

Afterwards you do this on the screenshot

[languid grove]

hello all, im new to this discord and would lke to ask how to instal the voice files on this server into so-vits-svc-fork to get a real time voice changer, can anyone lend a hand?

[pastel oak]

gptsovits dont work with realtime wokada

[rare goblet]

Ayo? @pastel oak level 55 !!!

[pastel oak]

only rvc models

[languid grove]

oh ok thank you, ill look into that right now

[unborn eagle]

Why I have this problem?

Ignoring faiss-cpu: markers 'sys_platform == "darwin"' don't match your environment
Collecting numba==0.56.4 (from -r requirements.txt (line 1))
  Using cached numba-0.56.4.tar.gz (2.4 MB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [8 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "C:\Users\wojte\AppData\Local\Temp\pip-install-3yy19jri\numba_299a783593934f0f9abf5b4ef7faca8b\setup.py", line 51, in <module>
          _guard_py_ver()
        File "C:\Users\wojte\AppData\Local\Temp\pip-install-3yy19jri\numba_299a783593934f0f9abf5b4ef7faca8b\setup.py", line 48, in _guard_py_ver
          raise RuntimeError(msg.format(cur_py, min_py, max_py))
      RuntimeError: Cannot install on Python version 3.11.4; only versions >=3.7,<3.11 are supported.
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

C:\Users\wojte\OneDrive\Dokumenty\RVC>```

[odd shale]

@pastel oak Can you help him?

[fiery raptor]

whats happening with applio

its been incredibly slow

[odd shale]

Just use Applio Colab.

[fiery raptor]

thats exactly what ive been doing

usually it takes it several seconds to convert my audios

but now it can take so long that the colab stops

[odd shale]

I have never gotten that issue.

[fiery raptor]

its never been this slow

[odd shale]

Maybe you should try splitting your audios in 2 parts?

[fiery raptor]

I'll try

ive uploaded longer audios for shorter times

but ill try

[odd shale]

Alright.

[fiery raptor]

its still slow

how weird

[odd shale]

I don't really know how to solve that..

[fiery raptor]

reminds me of the early ilaria huggingface rvc

those things would take hours to load

oh wait i think something's happening

[knotty moth]

if you have done global package install at another project/application, yeah it tends to cause conflicts

[unborn eagle]

can you help me with this?

[knotty moth]

use the precompiled installation and you won't have to do that hassle

[unborn eagle]

How to do it

[proven hill]

now it uses a gpu so its the fastest option on the market

[unborn eagle]

??

[rare goblet]

Ayo? @unborn eagle level 1 !!!

[unborn eagle]

.

[fallow crown]

@steel forge i need help with rvc

[rare goblet]

Ayo? @fallow crown level 1 !!!

[regal temple]

Does Mangio run on linux?

[fallow crown]

hey do u know how to install rvc

im having trouble

[last cove]

Every body i have one question

English or Spanish

Whoever type First is gay

[violet heron]

Haiii :3

[violet heron]

It can but mainline is recommended over Mangio

[pastel oak]

Doesnt work inside OneDrive where you put everything

You have python 3.11, you need to downgrade to 3.10 to make it work

[unborn eagle]

How to do it

[pastel oak]

google and download any 3.10.x version

[odd shale]

🐢: A

[regal temple]

Gotcha, are there any good guides for setting that up in Linux?

[rare goblet]

Ayo? @regal temple level 2 !!!

[unborn eagle]

Do I need to uninstall to install version 3.10.x?

[frozen monolith]

pls help

[pastel oak]

Yes

[frozen monolith]

help pls

[misty elk]

-hf

[azure marsh]

<:huggingface:1179800228946268270> Hugginface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by Ilaria Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -colabs, -kaggle, -local, -overtraining, -realtime, -rvc, -help

[finite pumice]

how to fix this

[unborn eagle]

C:\Users\wojte\Downloads\RVC>python -m pip install -U pip setuptools wheel
Nie mo
Why?

[last cove]

HE YOU TYPE FIRST. YOU ARE GAAAYYYY

[unborn eagle]

C:\Users\wojte\Downloads\RVC>pip install -r requirements.txt
'pip' is not recognized as an internal or external command,
operable program or batch file.

[violet heron]

Yayyyy :3

[violet heron]

your a Linux user do it yourself

https://cdn.discordapp.com/emojis/784338955759517696.webp?size=48

I think you should be able to download the version depending on your GPU and open the sh file inside

[regal temple]

What if i told you I have no idea what I’m doing in linux but my pipeline necessitates it

[frozen monolith]

pls help

when i run rvc it crashes everytime i try to convert

[timid valve]

technical question because i'm curious.. how are accents stored in the index file? and how does it extract them so fast? it only takes seconds to train the index file whereas it can take several hours to train the main pth, depending on the size of the dataset

[finite pumice]

@acoustic scarab what should i put the vac in? input or output?

[acoustic scarab]

out

vac in goes to out, vac out goes to in

[formal tartan]

while i'm talking it's sounds like robot how can i fix this i want make it closest to girl voice

[analog obsidian]

index doesn't really store the accent of the model, the accent is in the model itself, what index does is to learn the characteristics of the audios, this can be how the speaker pronounces certain consonants or how the speaker talks, so for example if you have a raspy voice model, decreasing index will decrease the raspy voice a bit but not fully remove it

so thats why is faster

[violet heron]

Think you should be able to unzip and run the sh file iirc

[timid valve]

ohhh that makes sense

super interesting

[violet heron]

Rename the “New Folder (2)” to something else without any spaces

[hot ledge]

guys m having problem with applio its said omegaconf 2.0.6 has a non-standard dependency specifier

does someone know how to fix it

[glad zealot]

Depends on your dataset

[analog obsidian]

values are more stable and the model sticks more to the dataset, leading to having less versatility
batch size is not really a quality setting, but a stability setting

8 or 16 for 7-10 minutes and above
4 for 5 minutes and below

however like hina said, it depends on* your dataset

16 is more accurate to the dataset

[austere tartan]

can someone come to the vc i am in and help me a lil bit setting up the a voice mod

[knotty moth]

nvm I suck at explaining that stuff

[indigo crater]

How do I prove that the model works?

[odd shale]

By testing it with any speech or singing audio.

Whenever you finish training a model, for testing it you can place any kind of audio sample (clean ofc with no instrumental or noise) on the "audios" folder, and then you put the path to the audio you wanna test on the Inference tab

[maiden remnant]

how do i downlaod rvc

i ahve inidia

[dense cape]

does rvc work with amd?
cuz i tried two version(MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.18a) and (vcclient_win_cuda_2.0.58-alpha) and they both said that i have an outdated cuda driver.

torch\nn\utils\weight_norm.py:28: UserWarning: torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.
warnings.warn("torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.")
2024-08-08 04:31:50.6181048 [E:onnxruntime:Default, cuda_call.cc:116 onnxruntime::CudaCall] CUDA failure 35: CUDA driver version is insufficient for CUDA runtime version ; GPU=-1364851804 ; hostname=ASUS2 ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc ; line=125 ; expr=cudaGetDeviceCount(&num_devices);
*************** EP Error ***************
EP Error D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc:125 onnxruntime::CUDAExecutionProviderInfo::FromProviderOptions [ONNXRuntimeError] : 1 : FAIL : provider_options_utils.h:153 onnxruntime::ProviderOptionsParser::Parse Failed to parse provider option "device_id": CUDA failure 35: CUDA driver version is insufficient for CUDA runtime version ; GPU=-1364851804 ; hostname=ASUS2 ; file=D:\a_work\1\s\onnxruntime\core\providers\cuda\cuda_execution_provider_info.cc ; line=125 ; expr=cudaGetDeviceCount(&num_devices);
when using ['CUDAExecutionProvider']
Falling back to ['CUDAExecutionProvider', 'CPUExecutionProvider'] and retrying.

[crude flame]

If you use zluda you can get training to work

[dense cape]

what is zluda?

[crude flame]

https://github.com/IAHispano/Applio/issues/503

[crude flame]

Cuda translation thing

[dense cape]

does amd work w-okada?

i tried it and also cuda error

[crude flame]

Yes

[dense cape]

(RVC1006AMD_Intel) i have this and t doesn't work

[rare goblet]

Ayo? @dense cape level 1 !!!

[crude flame]

https://github.com/deiteris/voice-changer/releases download the newest dml version

[dense cape]

i have an rx 6600

[dense cape]

hmm ok will try

[regal temple]

I am so sorry for being such a nub, what .sh file should I be running on linux in the mainline distribution in order to be able to run RVC?

[dense cape]

do i install this with zluda?

[crude flame]

No

You only need zluda if you want to train a model

[dense cape]

oh ok thank you ❤️

[proper mesa]

i just downloaded the voice changer and i can hear myself fine but it wont work in game or anything like that and i am not sure how to get it to work can anyone help me?

[dense cape]

thx it worked

[regal temple]

I'm getting this error when trying to run RVC in linux:

run.sh: line 54: ./tools/dlmodels.sh: Permission denied

What could be causing this?

[violet heron]

-realtime read the stuff on the voice changer guide about the audio cable

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[merry eagle]

how to fix this error
when i press download button i am getting this

[solar thunder]

Help me to dowload the whole thing
i have a macbook
im so confused

[violet heron]

Download what thing

[solar thunder]

the whole thing

i dont even know how to start it

or where

[violet heron]

What thing

Like let’s say you went a restaurant, and you ask for food

They don’t know what food you want

[solar thunder]

ai

[rare goblet]

Ayo? @solar thunder level 1 !!!

[solar thunder]

from the github

[violet heron]

Which ai from the GitHub?

[solar thunder]

i was trying to follow this https://youtu.be/We5oYpCR3WQ?si=PClqiDF07NlBRdA0 tut

but he has a window

and i have mac

and im lost

[violet heron]

-realtime

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[hidden crow]

use bootcamp

on mac

it gives u windows

[solar thunder]

i tried before

[hidden crow]

search up how to get bootcamp on mac

[solar thunder]

it didnt work

im gonna buy windows soon

[hidden crow]

u have to do alot of stuff though ts hard gng

[solar thunder]

so i just left it as it is

[hidden crow]

😭

[violet heron]

https://github.com/deiteris/voice-changer/releases/download/b2271/voice-changer-macos-arm64-cpu.tar.gz use this if your on a apple silicon mac

(Apple silicon is like M1/M2/M3)

[solar thunder]

it says

potential

damage

careful

[violet heron]

Discord gives that alert on any direct download link

[hidden crow]

@violet heron whats download link for windows?

[solar thunder]

oh

[violet heron]

What’s your GPU?

[hidden crow]

amd ryzen

[solar thunder]

imma just give up

[violet heron]

I said GPU, not CPU

[solar thunder]

bye

[hidden crow]

oh

[solar thunder]

thanks for trying to help

[hidden crow]

uhh lemme see

[violet heron]

Mac users when they have to do something that isn’t dragging the file to the applications folder

https://media.discordapp.net/stickers/1226512060477734992.png?size=160

[solar thunder]

i gave up

i like windows a lot

[violet heron]

What part do you give up on

[solar thunder]

everything of it

[violet heron]

Why tf do you have a mac

[solar thunder]

i bought this mac when i was 8

idk why i choose this

i could have gotten windows

[violet heron]

Click on the file to download it, unzip it, run the sh file

[solar thunder]

MY ADOBE STUFF COULD HAVE BEEN EASY THOSE SCHOOL STUFF

idk why

i bought

this damn mac

[violet heron]

Unless your mac has a Intel chip then your basically screwed

[solar thunder]

i have m2

chip

[hidden crow]

@violet heron radeon pro 555x

[rare goblet]

Ayo? @hidden crow level 1 !!!

[hidden crow]

is my gpu

[violet heron]

The M2 chip first released in 2022 and you said you got it when your 8, this means your now 10

Which means your underage

https://media.discordapp.net/stickers/1225910929619681382.png?size=160

[solar thunder]

idk

this

that was 2 years

ago

my memeory

si not memeorying

[solar thunder]

im not 10 certainly

[rare goblet]

Ayo? @solar thunder level 2 !!!

[solar thunder]

if i was i would not even know these

[violet heron]

-realtime might or might not work on yours, read the guide for the fork

[azure marsh]

This interaction has expired, use the command /guides realtime if you wish to see it again.

[violet heron]

if you were you would be watching duckus

https://cdn.discordapp.com/emojis/1163905903108554752.gif?size=48

[solar thunder]

damn

[hidden crow]

deadass

[solar thunder]

no

im not 8

[hidden crow]

deadass

[solar thunder]

no

im 13

with a short term memory loss

[violet heron]

:3

[hidden crow]

alr bro

[solar thunder]

that okay

i have my passport

to verify

again

i been banned on this acc a lot

[violet heron]

I wonder why

[solar thunder]

all for the same stuff

but i verified it

with my passport

[final sentinel]

I have edited the audio in bandlab after extracting the audio with Uvr5 ui and changing the voice with ilaliarvc, but this time there is a definite discrepancy, what should I do?

[knotty moth]

explain what kind of issue it is
どのような問題なのか説明する

[final sentinel]

When I put two audio data together in bandlab(daw), one with vocals only and the other without vocals, there is definitely a discrepancy between the two audio files.

It looks like this.

[knotty moth]

misaligned tempo, right? or you should probably do proper mixing
テンポがずれてるよね？あるいは適切なミキシングを行う必要があるでしょう

[pastel oak]

I dont know, at this point download a prepackage instead of manual idk

[pastel oak]

Set (MME) at the end for input and output, you selected two different types

any other stuff on rvc realtime check here

https://rentry.co/RVCRealtimeGuide

[brittle wing]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[rare goblet]

Ayo? @brittle wing level 2 !!!

[opaque spoke]

how do i start training models?

[rare goblet]

Ayo? @opaque spoke level 1 !!!

[pastel oak]

for local you need an nvidia RTX gpu

else you can use a cloud option

https://docs.aihub.wtf for guides and instructions

If you need cloud option, check pinned message and open Ilaria RVC the one for training, thats also a good one to use for training

[opaque spoke]

ok thanks

im using nvidia

holy that download is huge

how long do you guys think it will take to make a model?

[knotty moth]

depend on your gpu spec

[opaque spoke]

like learning

how to do it

[knotty moth]

https://docs.aihub.wtf/

[opaque spoke]

how long do you think it will take to learn

[knotty moth]

only God knows till you try

[opaque spoke]

how long does it normally take??

[rare goblet]

Ayo? @opaque spoke level 2 !!!

[opaque spoke]

i was just wondering how much time I would need to dedicate for this

[knotty moth]

you couldn't even predict tomorrow's weather accurately

[pastel oak]

Learning to clean a dataset is more important which the basics are explained and you can use right away
Training you do with a few clicks, the actual training process you monitor over with Tensorboard which is explained there too. training can take a few hours

You dont have to do anything while it trains

[main tinsel]

Why my training go back to epoch 1????

qte14_t3 | epoch=10 | step=3800 | time=20:27:54 | training_speed=0:02:16 | lowest_value=15.088 (epoch 10 and step 3687)
Training has been successfully completed with 10 epoch, 3800 steps and 34.271 loss gen.
Lowest generator loss: 15.088 at epoch 10, step 3687
Saved model 'C:\Users\dmg03\Downloads\Applio-main\logs\qte14_t3\qte14_t3_10e_3800s.pth' (epoch 10 and step 3800)
Successfully synchronized graphs!
Starting training...
Loaded pretrained (G) 'rvc\pretraineds\pretraineds_custom\G_KLM42_T4_40k.pth'
Loaded pretrained (D) 'rvc\pretraineds\pretraineds_custom\D_KLM42_T4_40k.pth'
0%| | 0/380 [00:00<?, ?it/s]C:\Users\dmg03\Downloads\Applio-main\env\lib\site-packages\torch\autograd_init_.py:251: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed. This is not an error, but may impair performance.
grad.sizes() = [64, 1, 4], strides() = [4, 1, 1]
bucket_view.sizes() = [64, 1, 4], strides() = [4, 4, 1] (Triggered internally at C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\reducer.cpp:334.)
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
qte14_t3 | epoch=1 | step=380 | time=20:30:29 | training_speed=0:02:13
qte14_t3 | epoch=2 | step=760 | time=20:32:38 | training_speed=0:02:08 | lowest_value=15.547 (epoch 2 and step 535)

This is what I reproduced. When I resume training, all D and G files will be deleted and go back to 1 epoch. crazy

[molten marlin]

Will the voice changer work on the rx?

570

[pastel oak]

yea use the fork

https://rentry.co/ForkVoiceChangerGuide

download amd Version

[molten marlin]

with no delay?

[pastel oak]

delay is always going to be happening especially on older cards

with rx 570 maybe 1 second not sure

[molten marlin]

thanks

[rare goblet]

Ayo? @molten marlin level 1 !!!

[molten marlin]

I will try

[brittle wing]

the voices are so choppy and not clear at all for some reason

[odd shale]

It's because some (if not the majority of public ones) models aren't trained properly or not suitable for realtime.

[knotty moth]

the D and G files were actually not saved before

[brittle wing]

even the 4 initial voices when I’ve downloaded the client

[rare goblet]

Ayo? @brittle wing level 1 !!!

[proper shale]

@brittle wing send your okada settings

let us see what's up

[odd shale]

Then it's probably something related to your okada settings.

Show your settings so MJ can help you.

[odd shale]

🐢

[proper shale]

hi my tortuguita

[odd shale]

Haiii my MJ

[proper shale]

how have u been

[proper shale]

[odd shale]

I've been doing fine..

Hiding like a turtle.

[proper shale]

i see

[molten marlin]

colab still working?

[odd shale]

I think there's a colab for W-Okada.

[molten marlin]

can i have it

[proper shale]

don't use crepe_full, use rmvpe instead

[brittle wing]

alr

everything else is fine?

[proper shale]

ig mess with chunk & extra to see what works best for you

and maybe use sup2 if u have BG noise

[brittle wing]

alr tq

[jagged marsh]

Hi how can I make the voice on w-okada sound like the samples? Any voice sounds nothing like the samples

[pastel oak]

Realtime works a bit different, you might have heard inferenced samples. So models may not all sound perfect on realtime

Best is to look for asmr, mommy, streamer, egirl voices for realtime.

dont forget to mess with the Tune / Pitch if its a different gender until it sounds right

[jagged marsh]

Yea I wanted to sound like Trevor Phillips lol

Or Obama, but the voice sounds like a white person

[pastel oak]

Maybe using index a bit can help

id guess Obama works fine since its mostly taken from speech

[analog obsidian]

Models not sounding like the original voice in realtime are because of multiple factors, but the main one is when your voice is too different from the original one, for example my voice is very high pitched and i talk very differently compared to the model i am using, this causing the model sounding weird despite being well trained

I noticed increasing index to 0.5 ish forces the model to pronounce consonants like the og voice and mimic more the way the original voice talks (still not perfect, the change is minimal, some models can do 0.7 index in realtime but not all),
but this can affect the model pronunciation (gets worse, it makes the model pronounce certain consonants like the original voice but can make even simple words have pronunciation issues) and sometimes even adds artifacts

So the best u can do for characters/known people is to actually mimic the way they speak
Sometimes the model may be undertrained, sadly u can’t fix that if u didn’t made the model

[pastel oak]

i love you lyery

[abstract vortex]

Hi guys, how long does this process usually take? Making an Ai cover btw

[rare goblet]

Ayo? @abstract vortex level 1 !!!

[odd shale]

I remember when i brought him to the staff team haha.

[odd shale]

It will depend on the index and pth's size

Pth file's will always only have 50+ MBs of size

But the .index size will always depend.

[abstract vortex]

Ohh okay, thank you! I havent made ai covers in a year and I was so used to using the easy gui haha everythings so different now

[odd shale]

Yep, EasyGUI is kinda outdated now

[brave garnet]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

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

Advanced Settings

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

⠀

Made with love by Ilaria 💖

[pale rock]

INSTALL_Mangio-RVC-v23.7.0_INFER where does this install what location?

[brave garnet]

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs
Version 18a cuda

Download for AMD GPUs
Version 18a directml

Download for Intel GPUs
Version 18a directml

Download for Mac
Version 17b Mac
⠀

Made with love by Ilaria 💖

[pastel oak]

Thats a prepackaged download, so all the files remain in that folder iirc

[brave garnet]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

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

Advanced Settings

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

⠀

Made with love by Ilaria 💖

[misty elk]

-kaggle

[azure marsh]

Kaggle Notebooks

• UVR5 NO UI by Eddy
• How to use RVC Mainline Kaggle by Cauthess
• 🍏 Applio RVC by IA Hispano
• ✨ RVC Mainline by Hina

More commands: -colabs, -hf, -local, -rvc, -help

[misty elk]

-overtraining

[azure marsh]

Overtraining

You can detect if a model is overtraining if the TensorBoard graph starts to rise and never comes back down. An overtrained model will sound robotic, muffled, and won't be able to articulate words well.

Check these resources to learn more about this topic

• Epochs & TensorBoard from AI HUB Docs
• TensorBoard from 🍏 Applio Docs

[abstract vortex]

using uvr and I keep getting this error?

here r my settings thingy

[brittle wing]

Cannot connect to GPU backend
You cannot currently connect to a GPU due to usage limits in Colab. Learn more
To get more access to GPUs, consider purchasing Colab compute units with Pay As You Go.

any way to fix without paying

[proven hill]

what you need?

[golden karma]

/guides uvr

-uvr

[azure marsh]

Ultimate Vocal Remover

One of the best free and open source vocal and instrumental isolation tool.

[pastel oak]

4h limit every day
make new google account if u want

[brittle wing]

what does rvc even mean

[pastel oak]

retrival based voice conversion

it does not mean real voice changer

[brittle wing]

hm

that’s enough for me tbh

[golden karma]

how to separate two voices in uvr?

[rare goblet]

Ayo? @golden karma level 2 !!!

[pastel oak]

You can try using UVR BVE on MVSEP (also for free) or one of the Karaoke models in UVR

[golden karma]

okay thanks

[radiant vortex]

Please wait

run with the runtime pythong

Or something

What the hell do I do=

[pastel oak]

What

[stiff helm]

How can I get RVC inference running locally on my Mac? I was using the Applio WebUI, but now I want to run it locally, without a WebUI. I want to be able to run it in my VSCode terminal. I just need to run inference, and be able to adjust the settings. I am familiar with Docker if I need to use that.

[sharp solstice]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[rare goblet]

Ayo? @sharp solstice level 1 !!!

[stiff helm]

AFAIK these are just for getting it to run on the WebUI

[stiff helm]

I want to be able to run it in VScode, without the webui

[violet heron]

RVC and Applio both have CLI versions

[regal temple]

w/ RVC AIO is it possible to train more than 1000 epochs?

[stiff helm]

I don't see on the Applio page where the have a CLI version (I don't think they do)

But I can check ine of the RVC repos

[vivid pewter]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[rare goblet]

Ayo? @vivid pewter level 1 !!!

[vivid pewter]

no module named gradio

why isnt it working

[timid valve]

stereo or mono? does it even make any difference?

[analog obsidian]

rvc converts the dataset to mono when preprocessing

[timid valve]

ah alright

[tidal trail]

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

[stiff helm]

Thank you, ill check this out

[stiff helm]

Well, there's no option for Mac

[vivid pewter]

-uvr

[azure marsh]

Ultimate Vocal Remover

One of the best free and open source vocal and instrumental isolation tool.

[analog obsidian]

@glad zealot this happen when trying to use mainline colab

[opaque spoke]

is there a guide on how you should merge models and stuff

[glad zealot]

link that you are using?

[analog obsidian]

https://colab.research.google.com/github/hinabl/RVC-Online/blob/main/Mainline_Colab_Full.ipynb

[glad zealot]

werd i just started it and it works no problem

[analog obsidian]

hmm, i'll try again now

[opaque spoke]

uh i have no clue what theyre saying lol

I'm like new and I was trying to see how I could make a realistic model

[open ivy]

need help with RVC, downloaded it and it shows me 6 options and wont show me any of the default voice models

[rare goblet]

Ayo? @stiff helm level 8 !!!

[analog obsidian]

yea its working fine now, idk i blame google lol

[glad zealot]

same

lol

[thin stump]

Take your best g/total and your best kl/mel for the merge. You upload the pth file on the folder here after you hit "download" and "run" https://colab.research.google.com/drive/1mHKTGH5e3SAyDSBss1KtiYRbDdQzwSMs

322 was best g/total and best mel/kl 442 epoch

[opaque spoke]

what is g/total and mel/kl?

[golden karma]

what's better crepe or rmvpe

[thin stump]

so read this first https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/#-setting-up. You can think of mel and kl being attached to each other. No need to overcomplicate it

and read this if you want #1213509354343637065 message

[rare goblet]

Ayo? @thin stump level 35 !!!

[analog obsidian]

for training or realtime conversion?

[golden karma]

realtime conversion

[analog obsidian]

rmvpe

[thin stump]

What I meant is that if kl rises on the graph while mel keeps going down, it's done. You can always wait it out to see if it changes

[opaque spoke]

oh ok ill try to figure it out along the way

[opaque spoke]

but if I want a better model should I train an ai model of my own voice and merge it with another model?

[thin stump]

It's just to improve clarity. I don't think you can take two different models and try to fuse that. We've already had someone fail miserably because he didn't listen to the errors saying you can't do that

[opaque spoke]

what do you recommend I should do?

[thin stump]

here #✨│ai-help message

first of all get started on your model first before you worry about tensorboards

[opaque spoke]

making my own voice model?

[thin stump]

Yup, we have guides. I think if you learn about what RVC likes, you'll get great results

https://docs.aihub.wtf/essentials/how-to-make-voice-models/
https://rentry.co/RVC-dataset-RX11

[golden karma]

what's the best model in uvr for separating harmonies?

[opaque spoke]

wait so I make a voice model of my voice and then I merge it with another?
or am I supposed to do something else?

[thin stump]

Make a dataset first > train in rvc > you have the .pth file and added_index. Do what I did here to merge pth files #✨│ai-help message

Again, it's a clarity boost for the overall model. But it wouldn't matter much if your dataset is noisy so please read the guides

[analog obsidian]

uvr bve tho is not perfect

[opaque spoke]

oh ok. is it fine if I ask which fork of rvc you recommend or should I just use the original? Also, how long should I record my dataset to be and how should I be speaking?

[thin stump]

Are you more of a local use person?

[analog obsidian]

merging a model of your real voice and a random model is not going to make it sound realistic, that is not how it works

[opaque spoke]

yes

[analog obsidian]

it will help it understand the way u speak more

but has the risks of sounding worse actually

[opaque spoke]

how should I improve the clarity then?

or is it not possible?

[analog obsidian]

realistic models are done using high quality audio, so no, if your current model is bad, you can't fix it

in realtime mic quality affects how clear your model sounds

and locally (rvc) the audio of the inference has to be clear as well

if your model doens't sound clear despite having both a good mic or a good audio, is a model problem

[thin stump]

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

^^ Record at least 10 minutes of lines or read an audio book or something. Too monotone and it'll sound monotone. You are the voice actor here

[opaque spoke]

cause like when I try to talk it slurs my words. That's probably because im using a laptop mic right?

[rare goblet]

Ayo? @opaque spoke level 3 !!!

[analog obsidian]

yes

[opaque spoke]

oh ok

hmmm

[analog obsidian]

basically for realtime you need a clear mic, low amount of background noise, and good volume mic

[opaque spoke]

so I should buy a mic?

do you have any reccomendations?

[analog obsidian]

first you have to try the model in rvc, locally, with a clean audio (can be a tts output, a clean raw vocal without effects, anything that is clear)

if its sounds clear there but not clear in realtime, is the mic

[opaque spoke]

how do you do that?

[analog obsidian]

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

follow the tutorial

[opaque spoke]

ok

thanks

[analog obsidian]

if your pc can't run the software, you can use a cloud solution (colab, kaggle)

but it should work
edit: btw i don't recommend training rvc models in a laptop, it can potentially overheat them, but inference should be fine (just converting audio files)

[opaque spoke]

do i need to have a good cpu and gpu?

[analog obsidian]

for inference? not really

for training? yes

[opaque spoke]

my laptop has a rtx 4060 and intel i7 13700h

would that be enough?

[analog obsidian]

yeah, but be careful about your laptop temperatures if you want to train with it

training is very intensive for laptops

[opaque spoke]

what should i do about it?

blow it with a fan???

or what

[analog obsidian]

idk

i just told u about that in case u notice high temperatures while training in your laptop

[opaque spoke]

like do i stop it if it gets too high or smth

[analog obsidian]

yea stop it, is very dangerous for your laptop to overheat

[opaque spoke]

can you have pauses in between training your model??

[analog obsidian]

yes, rvc saves the process, you can stop it and continue later

[opaque spoke]

oh ok

[vestal cradle]

I'm trying to integrate rvc into a python script for tts I see this https://github.com/blaisewf/rvc-cli and it seems to be what I need but I can not seem to get it to work in either Linux or Windows. I want it to do the tts and inference from inside of a python script and generate and output.wav does anyone have any experience with this or an implementation of RVC that can do this?

[dreamy berry]

how do i run beatrice v2? i downloaded the zip and ngl idk what to do next

[hushed steppe]

Can someone explain this

I've tried it in mp3 and wav it won't work
Even putting the name in lower case then reuploading it

Tag me if anything

[round quartz]

I'm currently using very high quality datasets to train models and i am wondering which pretrains would be the best to use? just the standard RVC V2?

[thin stump]

Still the OG as long as you are selective about what you choose for your dataset (making it clean). Combine that with graph syncing, fp32 training, model fusion = good model

AIhub were all about pretrains just for the clarity boost and values during training. So the pendulum has switched around because at the end of the day, Hifigan is crap

[round quartz]

Thank you! What is fp32 training and graph syncing? I have'nt heard of this.

[rare goblet]

Ayo? @round quartz level 1 !!!

[vestal cradle]

I've trained several convincing sounding models but i still haven't figured out how to read the tensorboard

[thin stump]

https://github.com/IAHispano/Applio/issues/405 also check the server logs

[final sentinel]

What is the setting for separating two or more hams in a hagginface uvr?

[hidden zodiac]

any fixes for voice cutting?

[odd shale]

You can read the W-Okada guides. (In fact, you can try and use Deiteris' fork)

[odd shale]

https://rentry.co/ForkVoiceChangerGuide

[final sentinel]

How do I set up Uvr5 ui to separate multiple hamori parts?

[odd shale]

Various harmonies?

You can try using the karaoke models.

Also, there's UVR online site which got Mel Karaoke.

[final sentinel]

メインボーカルを除く複数のバックボーカルがある設定を教えてくれませんか❓

[odd shale]

よくわからないから、UVRにあるカラオケのモデルで試してみてほしい。

[knotty moth]

カラオケ モデルは、通常中央にあるメイン ボーカルの抽出に効果的です。
バック ボーカルをさらに分離できない場合は、ポリフォニック検出モードで Melodyne を使用して分離してみることもできますが、品質が保証されない可能性があります。

[misty elk]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[tame wharf]

hi

i have this error in google colab

can't connect to gpu backend

[pseudo flint]

best settings for uvr5

roformer models included beta

[odd shale]

That error appears because you've already used the free hours Colab gives you for usage.

You can solve it by just using an alt google account

[tough fiber]

~paperspace

-paperspace

[azure marsh]

RVC v2 Tutorial for Paperspace

💡 Click on the button below to open the guide.

Credits: LollenApe

[tame wharf]

thank you

[odd shale]

Use Kaggle or Colab instead, RVC is right now unsupported on Paperspace since months ago

[rare goblet]

Ayo? @tame wharf level 4 !!!

[tough fiber]

@wispy lodge
yoo emoji
do u know how can i setup applio or mangio on paperspace
https://docs.google.com/document/d/1lIAK4Y0ylash_1M2UTTL_tfA3_mEzP0D2kjX2A3rfSY/edit
this guide doesnt work anymore i think

[wispy lodge]

That I dunno, I don't use paperspace

[tough fiber]

oh

[tough fiber]

need new guide really this one i think trash totally

[odd shale]

Use kaggle instead.

RVC is incompatible with Paperspace since months ago.

[tough fiber]

lol i paid just now

paperspace

they do refund or?

[odd shale]

I'm not sure if you can refund..

[tough fiber]

-kaggle

[azure marsh]

Kaggle Notebooks

• UVR5 NO UI by Eddy
• How to use RVC Mainline Kaggle by Cauthess
• 🍏 Applio RVC by IA Hispano
• ✨ RVC Mainline by Hina

More commands: -colabs, -hf, -local, -rvc, -help

[odd shale]

Read that guide made by Cauthess.

[tough fiber]

bro it says ngrok its too annoying thing

[odd shale]

Yep, you need to create a ngrok account.

[tough fiber]

yea i need to create every few hours or something

making a model is annoying

ill try on my local

really

[odd shale]

Uhh... Alright then.

[tough fiber]

i need place like paperspace

[odd shale]

Idk if you can run RVC on runpod then.

Or vast.ai

But there are no up to date tutorials or notebooks for running RVC on runpod or vast.

[tough fiber]

-applio

[azure marsh]

🖥️ Local stuff

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

Credits to Faze Masta and Antasma for compiling these links.

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

[tough fiber]

@odd shale i want to do perfect voice model. i have 3 hours good data
How long should the data last for best efficiency?

[odd shale]

Ehhh.. Just use 20 or 25 mins.

[proud pecan]

Hey Guys, what does this error in the colab version mean??

[odd shale]

It's not mandatory to use various hours.

[quick jungle]

https://cdn.discordapp.com/attachments/1160576561636966491/1165372018414342284/warning.gif

[proud pecan]

Building wheel for gin (setup.py) ... done
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
xgboost 2.1.1 requires nvidia-nccl-cu12; platform_system == "Linux" and platform_machine != "aarch64", which is not installed.
albucore 0.0.13 requires numpy<2,>=1.24.4, but you have numpy 1.23.5 which is incompatible.
albumentations 1.4.13 requires numpy>=1.24.4, but you have numpy 1.23.5 which is incompatible.
albumentations 1.4.13 requires pydantic>=2.7.0, but you have pydantic 1.10.17 which is incompatible.
chex 0.1.86 requires numpy>=1.24.1, but you have numpy 1.23.5 which is incompatible.
pandas-stubs 2.1.4.231227 requires numpy>=1.26.0; python_version < "3.13", but you have numpy 1.23.5 which is incompatible.
torchtext 0.18.0 requires torch>=2.3.0, but you have torch 2.0.1 which is incompatible.
torchvision 0.18.1+cu121 requires torch==2.3.1, but you have torch 2.0.1 which is incompatible.

Successfully installed all packages!

[pastel oak]

Broken, use different Colab

[proud pecan]

ok

[pastel oak]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[pastel oak]

or check pinned message for ilaria stuff

they good

[proud pecan]

maybe i shoul up date my things too 😅

[proud pecan]

ok thx

[zinc kraken]

Can someone help me I'm using local applio to train models. Does training take up a lot of space?

@odd shale ,

[low shard]

you finished your daily limited google colab gpu, you can:

• You can use an alt google acc
• Use kaggle that gives 30 hours weekly of more gpus
• Wait till tmr
• Pay for colab

[pastel oak]

i was p sure that errorlog was aicovergen being broken

Guess i was wrong

[tough fiber]

which one should i choose

custom?

[pastel oak]

dont ask me

[tough fiber]

well okay

[odd shale]

contentvec.

Don't change it

[odd shale]

Shad, if someone makes you a question about which embedder to use, just say "don't alter it, keep it on contentvec"

[pastel oak]

I had a feeling default is probably correct, but didnt want to give false info in case, thx

[odd shale]

Ur welcome Shady, and yep, contentvec is the default one

[low shard]

You could try https://docs.google.com/document/d/1ooG2hJrfNNLUln0reTKKIOpNBjp53G0joak50H_sQhE/edit that is made by a user here, but no one really uses paperspace here anymore so i dont think anyon can help you if it goes wrong

or also https://docs.google.com/document/d/1cvqC9k6rDpEqTxoCIJ274PCN3a7hnMlOAl1lQGgDy8M/edit

made by @plush dew , but idk if they still work

[plush dew]

i forgot about this lmao

if anyone still uses it i'll update

[upbeat cypress]

is there a specific program that you can use to use RVC models on silly tavern. I just have been using alltalk but was curious if I could actually load the RVC models since I know the extras api is been depracated.

[vestal cradle]

Is anyone familiar with https://github.com/blaisewf/rvc-cli I'm building a voice assistant python script that reads the response from openais api. I got it to do inference after modifying a function call in the rvc_cli.py but I never managed to get it to do tts. Are there any options for RVC TTS that can be called without loading a GUI?

[low shard]

is it broken?

[low shard]

alltalk is a different tts thing, rvc is speech to speech, idk if silly tavern got anything about rvc

i thought silly tavern was mostly about llms and tts

[upbeat cypress]

Yeah your correct. looks like the alltalk beta might have some rvc conversion though.

[plush dew]

idk i haven't used it in a while

[low shard]

RIP

u still use paperspace?

[low shard]

wait what?

[upbeat cypress]

https://github.com/erew123/alltalk_tts/tree/alltalkbeta

checking it out now

[inland cobalt]

does anyone know good settings for a women? (ex: index rate: 9)

[low shard]

never knew about this damn

seems interesting

[upbeat cypress]

seems to work towards improving the xtts model if nothing else.

[proven hill]

thanks!

[proven hill]

index rate has nothing to do with being a woman or a man hun

[pastel oak]

your stuff is literally best working on this server so ofc we recommend yours first

[brittle wing]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[proven hill]

i made it idiotproof and futureproof

2 months and no need to fix or update

[proud goblet]

I installed RVC, but this is what I get:
**Launching Retrieval-based Voice Conversion WebUI...

C:\Users\pozde\anaconda3\envs\rvc\python.exe: can't open file 'C:\WINDOWS\system32\Retrieval-based-Voice-Conversion-W ebUI\infer-web.py': [Errno 2] No such file or directory

TroubleChute One-Line installer:

(rvc) PS C:\WINDOWS\system32\Retrieval-based-Voice-Conversion-WebUI>**
what should I do to be able to launch it?

[rare goblet]

Ayo? @proud goblet level 2 !!!

[odd shale]

Ilaria is goated.

(you)

[worthy tulip]

sorry to ask but does anyone have the working rvc v2 disconnected colab the one i used to use is now offline

[naive spire]

Tiraram minha permissão de falar

[rare goblet]

Ayo? @quiet cairn level 1 !!!

[timid valve]

what training algorithm tolerates bad audio quality best?

i dont even want it to necessarily improve audio quality

i just want to know which one replicates it best

[analog obsidian]

are u talking about pitch extraction methods? they don't matter in how similar the voice will sound

[timid valve]

oh they don't? alright

do you advise against using custom pretrains?

aren't those just an extension of the OG pretrain?

[analog obsidian]

rmvpe is more robust (clear) and is more pitch accurate
mangio-crepe is less robust (less clear) and is slightly less accurate (by a very small amount, like 1% less), capturates more details than rmvpe, but is highly sensitive to noise

[analog obsidian]

yes

original already does everything

[timid valve]

"noise" as in white noise? or just poor audio quality? like the one you would get from a poor microphone

[analog obsidian]

as white noise yeah

[timid valve]

oh alright

[analog obsidian]

but also can be any type of noise, mouse clicking, a sfx residual, everything

mangio is very sensitive to that

[timid valve]

i see

[analog obsidian]

rmvpe handles those better

[proud anvil]

Please drop the google colab that works

[proud anvil]

Neural network

[analog obsidian]

but since it capturates more details about your audio, the result is very smooth and, sometimes, realistic

unlike rmvpe which likes to make your model sound a bit metallic/robotic

[timid valve]

oh okay

that all makes so much more sense now

thank you for the fast responses btw, you are always such a big help

like genuinely, i mean it

[analog obsidian]

no problem, in simple terms, if you want to capturate every detail about your audio, use mangio, making sure no residual noise is present

noise sensitive means is gonna add that specific sound in the whole model no matter what

hifigan is very bad at cloning noise profile sadly

[timid valve]

got it 👍

[rare goblet]

Ayo? @timid valve level 7 !!!

[timid valve]

anyway, i used to be a pretty well-known modder in some other community and dealing with a lot of questions could really get on your nerves, especially when people weren't very specific about what they wanted to know or when there was a language barrier

so i really appreciate what you're doing

and your patience

what are the downsides to setting the hop length as low as possible? since that does make it more pitch accurate, allegedly?

[proud anvil]

Please drop Google colab, a neural network, to create covers that will work

[tame mica]

-hf

[azure marsh]

Suggestions for @proud anvil

<:huggingface:1179800228946268270> Hugginface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by Ilaria Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces
• AICoverGen, by r3gm Huggingface Spaces
• Advanced RVC Inference, by r3gm Huggingface Spaces
• RVC v2 Huggingface version, by Clebersla Huggingface Spaces

More commands: -audio, -colabs, -kaggle, -local, -overtraining, -realtime, -rvc, -help

[tame mica]

second one

whar

[proud anvil]

Thank you very much

[timid valve]

besides the fact that it takes more time to infer

could it by any chance affect the quality of the model in a negative manner?

[analog obsidian]

crepe was trained at a hop length of 10 so anything below that may cause issues

[timid valve]

ahh

im going to train it on a hop-length of 32 for now since that's already half of what the default is

[analog obsidian]

oh wait sorry i got confused, crepe was trained with a hop length of 160

setting the hop length below that is fine xD

however i remember below 64 you can't really hear the change

[timid valve]

oh okay

[analog obsidian]

is not bad to set it below 64
i even trained with a hop length of 8 just fine xD

[timid valve]

it's just finished inferring so ig it doesn't matter at this point

[timid valve]

damn

that must've taken forever

[analog obsidian]

for speech 64-32 is good, below that i noticed is more prone to have voice cracks

[timid valve]

ooohh

👍

[analog obsidian]

remember that crepe (mangio) is less pitch accurate than rmvpe, so is not very good at speech

[knotty moth]

for inference, rmvpe is generally better

[analog obsidian]

yea because is more pitch accurate

so less voice cracks and notes wrong

[knotty moth]

imagine modding rvc

[analog obsidian]

also remember that inferencing a model in crepe is not going to add the benefits of crepe (removing the metallic sound of rmvpe)

to do that you have to train a mangio model and inference mangio

then you can notice the change

rmvpe models can inference crepe but ofc they perfom better inferencing rmvpe

they simply don't get the benefits of crepe

[timid valve]

so i should just use rmvpe when inferencing no matter what

[analog obsidian]

if your model is trained with rmvpe? yes, is better, more pitch accurate

mangio models sound better inferencing mangio, not because is more pitch accurate for them but because the audio get more smooth for them

[timid valve]

that makes sense

[knotty moth]

for the voice changer purpose, stick on rmvpe anyway

[analog obsidian]

yes, always

speech models always have to use rmvpe, mangio perfoms poorly on speech

asmr voices, soft voices can hide that fact tho

[timid valve]

yea i've noticed a lot of pitch inconsistencies

completely unrelated bujt something i noticed is how much faster training is once i disable the webui by turning off my browser..like pretty much twice as fast than with the webUI running

im using firefox and it eats up so much ram

[analog obsidian]

are u using mainline or applio

[timid valve]

it shoouuuld be main line

but i'm not 100% certain

how do i know?

[analog obsidian]

applio literally says applio and the ui is black and green

[timid valve]

ah no main line hten

[analog obsidian]

mainline has some chinese text in the training tab

[timid valve]

yes

main line then

which one's better anyhow?

[analog obsidian]

mainline

[timid valve]

i'm on the right path for once then

yay

[analog obsidian]

yea is true that closing the webui can save some resources

[timid valve]

okay so i'm not just imagining things xd

good

why are so many models roughly the same size, even when trained on much bigger data sets? they all seem to be around 54MB for me

i once trained a refined model with twice the amount of data than the first model and the size of model stayed roughly the same

[knotty moth]

same model parameters for the same architecture
(I mean take analogy of the billion param llm)

[timid valve]

oh so that's what mostly determines what size the model will have?

[knotty moth]

index file size

[timid valve]

yea that's where i noticed the highest discrepancies

[timid valve]

i had to look that up ngl but it's starting to make sense

once again, thanks for all the help, i learn so much from you

[proud anvil]

Please drop a google colab that will work, a neural network

[hybrid wagon]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[ocean quail]

I got Error
"'NoneType' object has no attribute 'setdefault on AiCoverGen

How do I fix it?

[pastel oak]

I believe AiCoverGen is broken atm. You should use something else

If its for inferencing, check pinned message for Ilarias RVC Zero, its fast and pre-set up on the huggingface space

You can also use command -colab and try out different ones.

[ocean quail]

I have problems with that as well

[pastel oak]

What kind of problems?

What did you open (link), and whats the error

[ocean quail]

GPU crash out something like that

[rare goblet]

Ayo? @ocean quail level 1 !!!

[final sentinel]

なんかずっとやってもエラーばっかりなんですけどどうすればいいですか？

[dim helm]

here it says error

[pastel oak]

Maybe it lost connection? It works for me, so perhaps restart, and then scroll back to top when you convert and check the top right for the updates on connecting to gpu etc.

[opaque spoke]

im trying to train a model. but where do i find the 32k sample rate option??

[pastel oak]

click on 48k, then back to 40k, it will appear

is bugged

[opaque spoke]

i tried but it still doesnt show

what should i do?

[tame mica]

click v1 then v2

[opaque spoke]

ohhhh ok thanks

is there a recommended settings to train my voice model with? is there like a guide or something?

like what do i put here?

[brittle wing]

1

[tame mica]

just keep most things default

[opaque spoke]

ok

[opaque spoke]

is this ouput good?

[proud pecan]

Hey Guys, does anyone know why cant I uplod models to the colab verion of the vc?

[rare goblet]

Ayo? @proud pecan level 1 !!!

[wanton bane]

there's Extraction type named rmvpe but i only got crepe and harvest in my RVC GUI
do i need to download rmvpe or need to download another version of rvc ?

[proud pecan]

Nvm it just takes a while

[proud pecan]

hmm good question, next question

[pastel oak]

i answred in #🧬│ai-chat

[proud pecan]

jk, but i sadly can´t help you wthat

[opaque spoke]

do you guys know what this error is?

[rare goblet]

Ayo? @opaque spoke level 4 !!!

[opaque spoke]

oh nvm

i am stupid

i didnt add extension

[final sentinel]

I keep trying and keep getting errors, what should I do?

[opaque spoke]

bruh nvm im still getting the error

[opaque spoke]

how long does it normally take to train a model with a 13 minute dataset

[knotty moth]

ただいまー！
GPUの使用制限に達しましたか?
よく分かりませんが、UVR関連の質問がある場合は、DMで送信するサーバー招待リンクにお気軽にご参加ください。

[proud pecan]

My Model is failing to import does someone know why?

[knotty moth]

failing to import in which rvc?

[proud pecan]

the Collab v2

[knotty moth]

what Collab v2

[proud pecan]

this one

[knotty moth]

if it were local okada, I would be able to answer

[rare goblet]

Ayo? @knotty moth level 28 !!!

[proud pecan]

uhhhh

what?

[pastel oak]

I dont recognize this, is this an old colab?

[proud pecan]

no a new one

[pastel oak]

Can you link it instead of screenshot

Else id recommend you check out this message with link and guide:
#🔍│help-w-okada message

[proud pecan]

yes here https://colab.research.google.com/github/w-okada/voice-changer/blob/v.2/w_okada's_Voice_Changer_version_2_x.ipynb#scrollTo=s7mYqKtW6VOI

[rare goblet]

Ayo? @proud pecan level 2 !!!

[proud pecan]

oh ok thx

[pastel oak]

ah this is colab of the new alpha version

First time seeing it for me 😭

[knotty moth]

ah I see

[proud pecan]

oh ok

[knotty moth]

you can instead use hina's colab, as well as the guide https://docs.google.com/document/d/e/2PACX-1vTIceEcBfS6Zqolv_QEysrFfI_EJikPxozWptP_EjkpLVl-l-gdo-ijBonQMTviAHEYm5emmd9k9TdC/pub

[proud pecan]

ok thx

[pastel oak]

the discord msg i linked links to that already

[proud pecan]

and the error is "ran out of input" btw

and how does this part work?

[pastel oak]

ignore that, it doesnt work

says so on the msg i linked

you upload models on the app when it opens

[proud pecan]

oh ok

thx again

[final sentinel]

I have exceeded the GPU utilization of the UVR5UI, what should I do? What can I do to avoid exceeding the gpu utilization as much as possible?

[knotty moth]

come back tomorrow or use a colab alternative
明日また来るか、colabの代替品を使用してください

[proud pecan]

i tried it like 15-20 times and it wont even start because it cant connect with the gpu, what should i do?

[low shard]

You can:

• Use alt google acc
• Use the ZeroGPU HF Space
• Wait till tmr
• Pay for colab

[low shard]

If you finished the Google colab GPU,
You can:

• Use alt google acc
• Use the Kaggle version (a bit harder than colab and requires a phone number)
• Wait till tmr
• Pay for colab

[knotty moth]

come back tomorrow if the gpu usage limit has reached

[steep mantle]

I want to know how to make a model and put it in the program

[rare goblet]

Ayo? @steep mantle level 1 !!!

[low shard]

What's your GPU?

[steep mantle]

NVIDIA

[final sentinel]

さっきまたuvr5やったんですけど設定変えたらできてこれってモデルが良くないってこちですよね？

[low shard]

Yeah but what is it? Rtx 3060?

[final sentinel]

Is there any setting other than karaoke model or uvr-bve that separates the background vocals from the main vocals?

[steep mantle]

It's rtx 1650

[low shard]

I think you mean GTX as all RTX models start with 2XXX

You can't train with that locally

You have to do it on cloud

For colab (4 hours of daily gpu for free, not much hours, but easy to use):

• Applio (ui)
• Mainline (UI)
• RVCDISCONNECTED (no ui)
  For kaggle (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
• Mainline (UI)

[knotty moth]

十分な品質が得られ、処理時間が長くなりすぎないように設定を調整できます。

[proud pecan]

ok tomorrow is ok

[cursive flare]

tysm now I know it

[odd shale]

Or you can just create alt Google accounts.

So you don't have to wait till tomorrow.

[hidden zodiac]

what does gpu0 gpu1 ect mean

[summer lantern]

i have one good working <333 after 4 months

[hidden zodiac]

i have issue with my own voice cutting any solutions?

[pastel oak]

You downloaded AMD version of the voice changer, there u have to select gpu number thats registered on your pc

[misty elk]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[teal dome]

will fcpe ever be fixed for applio

it was really useful for pronunciation differences

[glad veldt]

how do you denoise an instrumental?

like this

[rare goblet]

Ayo? @glad veldt level 5 !!!

[worldly totem]

where i can get onnx voice?

help me someone pls

[pastel oak]

You have to convert it to onnx yourself.

On w-okada voice changer you can do this by uploading the pth model, then clicking on export to onnx

[rare goblet]

Ayo? @worldly totem level 3 !!!

[worldly totem]

I can, but it doesn’t work, it just wheezing

[pastel oak]

What do you mean wheezing

Whats the problem rn why do you need onnx

[worldly totem]

like white noise

[worldly totem]

The onnx voices I make myself just produce white noise

[pastel oak]

I dont know why it does that but why do you need onnx anyway? for what

Maybe theres an alternative

[grave mortar]

Hi everyone! I have a problem installing fairseq through pip. Does anybody has the same issue

[knotty moth]

are you talking about colab notebooks? try downgrading pip to ver 23

[grave mortar]

how do i downgrade pip?

[unborn geyser]

where i can get snowie? pretrain

[worldly totem]

I think the problem is in onnx which my computer makes, maybe onnx ready voice will fix my situation

[pastel oak]

!pip install pip==23.1

[pastel oak]

That didnt answer the question again

Anyway people dont upload ready-onnx files

[glad veldt]

Any help?

[wild vapor]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[rare goblet]

Ayo? @wild vapor level 2 !!!

[grave mortar]

thank you so much, i ve been looking for a solution a very olng time

[rare goblet]

Ayo? @grave mortar level 1 !!!

[grave mortar]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• Full AI Voice Model Training Guide (local) by Christopher Villanueva
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[hasty crow]

-audio

[azure marsh]

📚 Audio Guides

• Creating Datasets for RVC using iZotope RX11, by Cauthess
• Perfecting Audio Isolation on Low-End Rigs, by Litsa The Dancer and Faze Masta
• Gathering and Isolating Audio, by SCRFilms :snowflake:
• Vocal Mixing Tutorial, by Roomie

🛠️ Tools

Audio Separation/Isolation

• https://mvsep.com/
• https://multimedia.easeus.com/vocal-remover/
• MDX-NET Colab

More commands: -colab, -uvr, -karafan, -overtrain, -help

[low shard]

you are prolly using the easygui google colab, reminder that its abandoned and there are other issues with it too

[grave mortar]

im using

one sec

[grave mortar]

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

why doesnt it detect my GPU?

[low shard]

are you doing it locally or on google colab?

[grave mortar]

locally, on my pc

[low shard]

Alr, https://docs.aihub.wtf/rvc/local/mainline/ could help you

[grave mortar]

u mean i just need to download RVC1006Nvidia.7z instead of mine?

or troubleshoot using the given link?

[odd shale]

Yep, that link contains things which will may help you.

[low shard]

yes

[grave mortar]

oh

[misty elk]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, modded by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
• Ilaria RVC, by thestingerx Google Colab
• UVR5 NO UI for Google Colab, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab

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

[grave mortar]

its working, thank u

[low shard]

Yw

[glacial rapids]

how do I calculate the processing time of a conversion? says 642/0.9 atm and still going / am running a local conversion

[naive spire]

They took away my permission to speak, bro

Just because I spoke my native language on a call that wasn't in my language

[narrow sentinel]

how do you use this https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=DZDKirCM0F9g

[proper shale]

honestly, you should use something else instead

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

better training colab

easygui is straightforward tho, if u wanna keep using that, but no guide for it

[narrow sentinel]

all im tryna do is get to gradio(whatever it's called) and find some like a text to speech

[proper shale]

oh, in that case just use ilaria rvc

[narrow sentinel]

can u send please

[proper shale]

https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit

guide + link

[narrow sentinel]

Tsym

[rare goblet]

Ayo? @narrow sentinel level 1 !!!

[proper shale]

you're welcome :)

[proper shale]

it can vary, but normally it doesn't take that long unless it's using your cpu

can you share your specs and what rvc you're using?

[glacial rapids]

the file being processed was 5 minutes long. ticked to ~830 before completing. running a 1060 6gb, 32gb of ram on a ryzen 3600 with a noctua aftermarket fan

im using whatever RVC was available on the ".wtf" guide for local

[proper shale]

hm... that may be the good ol' 1060 screwing you up

[proper shale]

there's 3

applio, mainline, mangio

[glacial rapids]

mainline

[rare goblet]

Ayo? @glacial rapids level 1 !!!

[proper shale]

mm

yeah that's probably the gpu in this case

[glacial rapids]

the 1060 does good enough work for me, ill give it another year before i need to upgrade. will probably scale the entire build up again

[gaunt moss]

my rvc app keeps crashing everytime i launch it

[viral crater]

Does anybody know how to fix
'No model for model_common loaded. Please confirm the model uploaded.'

[gaunt moss]

like something about tmp

[pastel oak]

which rvc app did you download, whats the name

[gaunt moss]

voice.ai

[pastel oak]

on which program is this happening

[pastel oak]

We dont work with voice.ai and dont recommend it, so we cant help tbh

[viral crater]

Realtime Voice Changer Client

[pastel oak]

on which step does this happen? Are you trying to upload a voice model?

[viral crater]

i uploaded the index file and pth file

but when i start the console says: 'No model for model_common loaded. Please confirm the model uploaded.'

[rare goblet]

Ayo? @viral crater level 1 !!!

[pastel oak]

Did you select the model aswell on the blue box?

[viral crater]

blue box?

[pastel oak]

where the voice model is uploaded in

you have to select the voice model after you uploaded it, is it selected?

[viral crater]

i see this in my program

[pastel oak]

Oh brother

this version is old

[viral crater]

hm?

oh

[pastel oak]

How do people keep downloading the old version lately haha

[viral crater]

because its still available ig

[pastel oak]

https://rentry.co/VoiceChangerGuide

has latest versions

[viral crater]

so do i just delete it and reinstall latest version?

[pastel oak]

if you have AMD gpu, download this instead
https://rentry.co/ForkVoiceChangerGuide

[pastel oak]

delete it, and open one of the two guides and use the download from there

[viral crater]

how to know what gpu i have?

[pastel oak]

Task manager > Performances

Check for gpu0 and, if it exists, gpu1 name

[viral crater]

uhh where gpu

im not sure i think i have nvidia tho

[pastel oak]

Is there anything else under "wi-fi"

if you scroll down

[viral crater]

no

[pastel oak]

Then you have no gpu

[viral crater]

O_O

[pastel oak]

CPU voice changing is not recommended, youll damage your pc in the long run probably if you play a game with it too and cant do much with it xd

If your internet is good, you can try the online hosted alternative

[wispy lodge]

There cannot be no gpu, there should be at least some gpu

[viral crater]

why the heck do i have no gpu O_O

[rare goblet]

Ayo? @viral crater level 2 !!!

[wispy lodge]

Could you check device manager maybe?

[pastel oak]

I assumed iGPU but now that you mention it, it would still show up separately

I forgot 💀

[viral crater]

ok i will just download nvidia ig

i have nvidia drivers so it makes sense to download nvidia

[pastel oak]

Yeaa

Did you check device manager tho?

[viral crater]

what section in dm?

[pastel oak]

Device Manager > Graphics Card

[viral crater]

._.

i dont have a gpu do i

[pastel oak]

You should have one

[viral crater]

ok then windows is broken lol

god does this program ever take long to download

[odd shale]

Yep, W-Okada got quite a size.

[viral crater]

5 gigs jesus

i use external drive

[grave mortar]

Sup everyone! Is there a program to create an rvc dataset for me woth given .wav files? All programs that i saw online is buggy/not working at all. Thanks in advance

[rare goblet]

Ayo? @grave mortar level 2 !!!

[viral crater]

help the client doesnt reopen

[pastel oak]

do this
https://cdn.discordapp.com/attachments/1159290161683767298/1159607580184821760/issue.mp4?ex=66b93137&is=66b7dfb7&hm=5b305eebcfad5a925c04947517303783f037674c93559e23dd0b6462a40940af&

[viral crater]

it says 'Web Server Launch Exception, Expecting value: line 1 column 1 (char 0)' and then never opens

[odd shale]

Nope.

The best way to make a dataset is manually

Gather at least 12-15-20 mins of dataset, split them on clips of 5-10 seconds and place them on a folder.

[pastel oak]

does it say a bunch of wait web servers after

[viral crater]

yes