✨│ai-help

[proven hill]

no matter if your input is stereo

[meager terrace]

I understand, and do you use smth thats makes your model not cap at 15Hz?

[proven hill]

not as im aware of

@brittle wing

[brittle wing]

Capping at 15Hz or KHz?

[glad zealot]

note on colab covergen ig.. its not gonna work with gpu atm since google updated stuff so gpu is acting weird with it

[meager terrace]

15KHz (A bit above 15 btw)

[finite galleon]

Did my colab that i sent you earlier this day had the same problem

[white bough]

Is there a way to fix it on our side, by changing the modules/requirements? The colabs that make models do not seem to be impacted by this, so it is probably just incompatibilities between the updated modules...

[glad zealot]

testing atm cant find yet

but kaggle work

[glad zealot]

https://www.kaggle.com/code/hinabl/multinotebook-ai-public you can try this, just set the rvc number to 4 since thats covergen

[white bough]

Thanks. Is there any guide to use kaggle? I've put RVC = "4". I ran the cells, but the "Set Up URLs" cell gives an error. Should I fill something else?

[rare goblet]

Ayo? @white bough level 2 !!!

[glad zealot]

are you fully registered?

look on the right side for this

make sure internet is on

[white bough]

Thanks!! Internet was not on. The cells are running now. Let's see if it works till the end!

[glad zealot]

make sure to set ngrok token

and have gpu enabled

[white bough]

Yeah, I'm seeing the ngrok problem now, but I'm managing atm. I will ping you if I have a problem. Thanks for your help!

[glad zealot]

yup feel free to ping

[stuck vessel]

How to train the model?

[latent kettle]

Use rvcDisconnected

-colab

[azure marsh]

All working colabs / spaces ☁️

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

Credits to Faze Masta and Antasma for compiling these links.

[latent kettle]

Third one

[stuck vessel]

What I have to do? I need guidance

[stuck vessel]

What I have to do? I need guidance

[rare goblet]

Ayo? @stuck vessel level 1 !!!

[latent kettle]

Firstly you have to prepare a dataset

[stuck vessel]

Character audio data?

[latent kettle]

Yes ! You have to collect data character voice. Make sure it should be clear and crispy!

[stuck vessel]

What do I need to do next?

[latent kettle]

Data should be of 10 minutes in length

And in (.wav) format

I recommend you to take 20-30 minutes of data for better output! Also you have to remove eco reverbe and long pause(s) from the audio file! You can Use UVR 5 to do that!

[indigo topaz]

my thing keeps getting stuck on seperating vocals from instrumental what should i do

[stuck vessel]

done

next?

[latent kettle]

Okay now make a .zip file

[latent kettle]

Are you using UVR 5?

[stuck vessel]

yeah

next

[indigo topaz]

whats that? im using AICoverGen-WebUI, modded by Hina Google Colab - whcih worked for me multiple times a couple days ago and now just keeps getting stuck eg.

wont let me send an image

[latent kettle]

UVR 5 is a kind of software used for separation of vocals from instruments or vice versa

[latent kettle]

Now open RVCdisconnected

[latent kettle]

Try to use it with alternative accounts !

[indigo topaz]

last time i did it, it seperated vocals and instrumentals for me and now it wont work, i even tried seperating my vocals and then doing it

[rare goblet]

Ayo? @indigo topaz level 1 !!!

[latent kettle]

AiCovergen is used to make AI covers!

[stuck vessel]

done

[indigo topaz]

im confused i uploaded a song i made and tried to use an ai cover voice, i did this before and it worked and now it just keeps getting stuck any idea on what i shoudl do please?

it gets stuck saying seperating vocals and instrumentals even if i just use vocals

so idk whats going on

[latent kettle]

Before starting upload your dataset zip into Google drive . Note you should Use same Google account in colab as well as in Google drive!

[stuck vessel]

i'm understand

[latent kettle]

It doesn't matter! You can upload vocals as well as a complete song! It will seperate it! Because it is programmed to do that! So if you think you will directly upload vocals! It's useless!

[indigo topaz]

i uploaded the complete song with both vocals and insturmentals and it didn't work! it keeps getting stuck trying to seperate it!!!!

[latent kettle]

I think it's runtime error! Check your runtime on colab!

[indigo topaz]

no error whenever i do it

i tried multiple times

[latent kettle]

now you are using webui?

[indigo topaz]

yes

[zealous shuttle]

is the colab run slowly? if so that's happening to me too

[stuck vessel]

next step @latent kettle

[obtuse herald]

hi
whats this problem?
Traceback (most recent call last):
File "D:\Folders\aishit\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\infer-web.py", line 316, in vc_single
audio_opt = vc.pipeline(
File "D:\Folders\aishit\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\vc_infer_pipeline.py", line 539, in pipeline
sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
RuntimeError: CUDA error: the launch timed out and was terminated
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

[latent kettle]

Go to colab! And check runtime is connected or disconnected

[indigo topaz]

well now i stopped it but I remember it was connected

[finite galleon]

aishit 🔥

[obtuse herald]

ye😂

[rare goblet]

Ayo? @obtuse herald level 1 !!!

[latent kettle]

Done with uploading? Now open RVCdisconnected notebook and run 1st cell dependencies

[latent kettle]

If you are not utilizing GPU or you are inactive ..it will automatically disconnected after some time

[patent pasture]

Is AICoverGen working?

[finite galleon]

Should be

[latent kettle]

Yes!

[patent pasture]

Just in past process took about 3 minutes but nor it has been lasting for 10 minuts already

[zealous shuttle]

can you send your colab link?

[patent pasture]

What have I to do?

1170 seconds and only 20% of progress

[fringe moon]

Hello. I can't find the Turkish content in this sound modification room in the cloning, i.e. sound models room. can you help me?

[obtuse herald]

anyone?

[stuck vessel]

it doesn't show anything, no code?

[rare goblet]

Ayo? @stuck vessel level 2 !!!

[latent kettle]

What?

[stuck vessel]

Dependencies just run

[latent kettle]

Have you mounted you Google drive?

[stuck vessel]

yeah

[patent pasture]

@glad zealot , hi, sorry but does AICoverGen work?

[latent kettle]

Yes it works! Check your runtime!

[latent kettle]

Now move 2nd cell

[stuck vessel]

Preprocessing?

[patent pasture]

I dont know is it normal? Process has been continuing for 1500 seconds and didn't end?

[sand iris]

Oh yeah

[finite galleon]

https://colab.research.google.com/drive/1X8YR4Ruv7zzY8YAMPTfC7hkxqT_d4Q5d it is normal inference no auto seperation

[sand iris]

The colab VMs recently updated

Making GPU support for AIcovergen break

for the most part

[patent pasture]

Oh, thank you

[sand iris]

Hina's working on a fix, unsure if it's actually doable

[zealous shuttle]

its okay, i laready separate it on uvr

[patent pasture]

thanks, I will try

[latent kettle]

Move to Set Training Variables And use these settings.. you can write your own model name in experiment_name avoid to use space " "

[patent pasture]

@finite galleon, shell not I change other and advanced settings?

[finite galleon]

If you know what your doing yes but if not then dont

[glad zealot]

broken atm google updated

[patent pasture]

ok

[stuck vessel]

and run?

[zealous shuttle]

guess we just waiting for colab updated then

[latent kettle]

Don't close your Google colab Notebook and open Google drive you will see a new folder 📂 named rvcDisconnected move your dataset in that folder

[glad zealot]

https://www.kaggle.com/code/hinabl/multinotebook-ai-public this works just make sure to set the rvc to 4 and your ngrok token

needs to have a verified kaggle account first tho

[patent pasture]

Oh, thank you guy

[zealous shuttle]

too lazy to use kaggle

[verbal oxide]

i'm worrying if kaggle blocked again

[glad zealot]

works

[verbal oxide]

yeah

[patent pasture]

covergen is soon or it is already there?

[glad zealot]

its there

set rvc to 4

[patent pasture]

ok, thank you so much!

Now I can create covers for 30 hours)

[glad zealot]

12* but ye

[stuck vessel]

@latent kettle it doesn't run

[latent kettle]

Don't close your Google colab Notebook and open Google drive you will see a new folder 📂 named rvcDisconnected move your dataset in that folder

[stuck vessel]

i don't close

but it doesn't run

[latent kettle]

DM me

Nd send me a screenshot

[latent kettle]

Change these settings

[latent kettle]

To this

[stuck vessel]

right?

[dire wagon]

dr jr can u help me ?

[latent kettle]

Yeh

[dire wagon]

i always get the tensorRt, pip install tensorboard thing

how can i fix it

[latent kettle]

Hey wait for few minutes. I'll send you a complete tutorial!

[latent kettle]

I don't know much about it! But you can run that command which it suggest!

[dire wagon]

doesnt work at all

[latent kettle]

Emm but sorry I don't have enough knowledge regarding it

[dire wagon]

can u send me a tut with him ?

[latent kettle]

No it's not for you, i don't know about your error

[pale beacon]

One message removed from a suspended account.

[dire wagon]

ik i dont know how to use rvc v2 too

so i need help me with it

[latent kettle]

Which notebook,

[molten pecan]

-huggingface follow the first two steps only :3

[azure marsh]

Suggestions for @pale beacon

huggingface.co upload Guides

English guide

Link: https://rentry.org/fdg_guide_new
Credits: FDG

Spanish Guide

Link: https://rentry.org/AI_Hub_guide_esp
Credits: Julia (ailen2091)

Hindi guide

Link: https://rentry.org/AI_hub_guide_hi
Credits: Enes

Turkish guide

Link: https://rentry.org/AI_hub_guide_tr
Credits: Enes

[molten pecan]

once you are done copy and paste the HF link to Colab

[pale beacon]

One message removed from a suspended account.

[zealous shuttle]

this one?

[glad zealot]

yers

[stuck vessel]

.

[zealous shuttle]

how to get a ngrok token?

[rare goblet]

Ayo? @zealous shuttle level 9 !!!

[glad zealot]

ngrok.com

also make sure you are verified (phone number)

[zealous shuttle]

already verified

[patent pasture]

it gave me an error - dial tcp 127.0.0.1:9303: connect: connection refused

[glad zealot]

try again

[patent pasture]

oh, it worked!

[zealous shuttle]

and where i can get the token? already register the ngrok account

[patent pasture]

you are our hero)

[glad zealot]

on the left side theres a part that says "your auth token" or soemthing

[zealous shuttle]

only setting this one? the rest is default?

[glad zealot]

if you're doing covergen so ye

you dont need to change anything

[zealous shuttle]

if i already had a model on hf, should i download the model and upload it on kaggle on model part (right side)?

[glad zealot]

nah you can just upload it in the gui later

on the download tab

[zealous shuttle]

when do a inference what accelerator should i use?

[glad zealot]

t4x2 ig

both p100 and t4 works fine

[zealous shuttle]

it must to run all cell first?

[glad zealot]

all cell

from top

if you skip one it wont work at all

[zealous shuttle]

click the covergen url?

[rare goblet]

Ayo? @cloud merlin level 3 !!!

[silver anvil]

stop

[glad zealot]

wut

[silver anvil]

@cloud merlin stop pinging

[zealous shuttle]

NVM found it

[cloud merlin]

@silver anvil @glad zealot @zealous shuttle anyone help

[sand iris]

lmfao

[molten pecan]

dude 😭

[sand iris]

gl lad

[silver anvil]

-howtoask

[azure marsh]

How to troubleshoot

• Give context: Throwing your error/issue alone will not help. Tell what step are you on, what were you doing before issue, your settings, your terminal, et cetera.
• Be relevant: You come to ask for help, not to complain. You can't ask for RVC help in voice changer help and vice versa.
• Be polite: DONT DEMAND FOR HELP OR I WILL FUCKING MURDER YOU. Ask politely, the server was not made for you.
• Don't be an attention whore: If people are ignoring your help, kindly ping them. If they are ignoring you, either they don't know how to solve it, they are busy, or you are an asshole.

WITH ALL DUE RESPECT, DO NOT BE AN ASSHOLE AND HAVE COMMON SENSE.

[silver anvil]

@cloud merlin@cloud merlin@cloud merlin@cloud merlin@cloud merlin

[molten pecan]

OH MY GOD

😭 😭

[sand iris]

@cloud merlin

https://cdn.discordapp.com/emojis/1093683459471458324.gif?size=48&name=clueless_ball&quality=lossless

[cloud merlin]

pls help

[molten pecan]

@cloud merlin how to make ai cover

[viscid moss]

Hello

`2023-12-15 14:40:04.791324012 [E:onnxruntime:Default, provider_bridge_ort.cc:1480 TryGetProviderInfo_CUDA] /onnxruntime_src/onnxruntime/core/session/provider_bridge_ort.cc:1193 onnxruntime::Provider& onnxruntime::ProviderLibrary::Get() [ONNXRuntimeError] : 1 : FAIL : Failed to load library libonnxruntime_providers_cuda.so with error: libcufft.so.10: cannot open shared object file: No such file or directory

2023-12-15 14:40:04.791366893 [W:onnxruntime:Default, onnxruntime_pybind_state.cc:747 CreateExecutionProviderInstance] Failed to create CUDAExecutionProvider. Please reference https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements to ensure all dependencies are met.`

Any clue how to fix that sht? I'm trying to fix AICoverGen too

[silver anvil]

-howtoask

[azure marsh]

How to troubleshoot

• Give context: Throwing your error/issue alone will not help. Tell what step are you on, what were you doing before issue, your settings, your terminal, et cetera.
• Be relevant: You come to ask for help, not to complain. You can't ask for RVC help in voice changer help and vice versa.
• Be polite: DONT DEMAND FOR HELP OR I WILL FUCKING MURDER YOU. Ask politely, the server was not made for you.
• Don't be an attention whore: If people are ignoring your help, kindly ping them. If they are ignoring you, either they don't know how to solve it, they are busy, or you are an asshole.

WITH ALL DUE RESPECT, DO NOT BE AN ASSHOLE AND HAVE COMMON SENSE.

[silver anvil]

^ read this @cloud merlin

[molten pecan]

@true vortex

[half cove]

@cloud merlin can u help me in fortnite?

[viscid moss]

[glad zealot]

no its dead atm

google updated

[zealous shuttle]

can it detect from local laptop?

[viscid moss]

ya ik

[glad zealot]

wot

[dense crane]

@cloud merlin how do I add voice model to Minecraft

[molten pecan]

yeah of course. you can even use it on mobile

[zealous shuttle]

on Song Output (acapella one)

[molten pecan]

if that was your question

[glad zealot]

your own audio from your laptop ?

[zealous shuttle]

yeah

[glad zealot]

now sure ngl i only used youtube

[viscid moss]

currently trying this:

!apt-get --purge remove "cublas*" "cuda*"
!apt-get --purge remove "nvidia*"
!dpkg -l | grep cuda- | awk '{print $2}' | xargs -I {} dpkg --purge {}
!dpkg -l | grep nvidia- | awk '{print $2}' | xargs -I {} dpkg --purge {}
!apt-get install cuda=10.0.130-1
!apt --fix-broken install

But still doesn't works

[glad zealot]

need to somehow downgrade cuda to 11(what it was before) since its 12 rn

im trying but couldnt make it work sadge

[viscid moss]

https://tenor.com/view/ah-shit-here-we-go-again-gta-sa-cj-gta-again-gif-20425950

[glad zealot]

!find / -name '*libcublasLt.so*' from gdr but ye you can check the version using this command

[viscid moss]

ty

[glad zealot]

this is on kaggle find: ‘/proc/35/task/35/net’: Invalid argument find: ‘/proc/35/net’: Invalid argument /opt/conda/pkgs/libcublas-11.11.3.6-0/lib/libcublasLt.so.11.11.3.6 /opt/conda/pkgs/libcublas-11.11.3.6-0/lib/libcublasLt.so.11 /opt/conda/pkgs/libcublas-dev-11.11.3.6-0/lib/stubs/libcublasLt.so /opt/conda/pkgs/libcublas-dev-11.11.3.6-0/lib/libcublasLt.so /opt/conda/pkgs/cudatoolkit-11.8.0-h4ba93d1_12/lib/libcublasLt.so.11.11.3.6 /opt/conda/pkgs/cudatoolkit-11.8.0-h4ba93d1_12/lib/libcublasLt.so /opt/conda/pkgs/cudatoolkit-11.8.0-h4ba93d1_12/lib/libcublasLt.so.11 its still on 11

[cloud merlin]

first help me then i provide u link

[viscid moss]

[cloud merlin]

@molten pecan

help

[viscid moss]

find: ‘/proc/57/task/57/net’: Invalid argument
find: ‘/proc/57/net’: Invalid argument
/usr/local/lib/python3.10/dist-packages/torch/lib/libcublasLt.so.11
/usr/local/cuda-12.2/targets/x86_64-linux/lib/stubs/libcublasLt.so
/usr/local/cuda-12.2/targets/x86_64-linux/lib/libcublasLt.so
/usr/local/cuda-12.2/targets/x86_64-linux/lib/libcublasLt.so.12
/usr/local/cuda-12.2/targets/x86_64-linux/lib/libcublasLt.so.12.2.5.6```

My colab output 💀

[glad zealot]

yes it needs to be 11

its 12

so ye..

[molten pecan]

no idea how to fix that. I don't use Colab

[viscid moss]

[glad zealot]

ditto

[molten pecan]

use my guide if you want https://rentry.org/ilarvc_inf_guide

[glad zealot]

i dont use rvc disconnected

[molten pecan]

ditto?

[cloud merlin]

@molten pecan https://huggingface.co/spaces/Clebersla/RVC_V2_Huggingface_Version

[glad zealot]

same\

[cloud merlin]

i use this

[molten pecan]

did you duplicate the space first?

[cloud merlin]

just used it as it was

[sand iris]

@glad zealot I believe you can fix cuda on aicovergen actually

cuda11 is available in the repos

[molten pecan]

click the three dots, Duplicate this Space, and then Duplicate

wait for a few minutes

[glad zealot]

hoooo

[molten pecan]

and then you are good to go

[glad zealot]

want

[sand iris]

The colab VMs are Ubuntu 22.04

https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/

the packages are there

cuda-11-8_11.8.0-1_amd64.deb

[glad zealot]

ye but how to put it on colab

[sand iris]

!sudo apt install cuda-11-8

and possibly other stuff

[glad zealot]

also i tried the one i send in dm and it took so long

[sand iris]

cba to check rn

[glad zealot]

cant really test rn going to bed

[sand iris]

oh my bad, sorry then lol

gn

[zealous shuttle]

so for example if there is still a accelerator quota, and i let it for 1 week does it reset again to 30 hours or not?

i've been using it for first 30 minutes

[sand iris]

sup

[glad zealot]

imma go save that tho gonna test it

[sand iris]

!sudo apt install cuda-11-8

add that to like the first code cell

[glad zealot]

yes it resets to 30 hours after a week

[sand iris]

and report back pls

[viscid moss]

okkie

[glad zealot]

@sand irisand@viscid mossgonna be in modified credits OwO

[cloud merlin]

@sand iris i am using this https://www.youtube.com/redirect?event=video_description&redir_token=QUFFLUhqbHduUlVkM0NkdVRvUms4bVBtVk5NSUR6VGppd3xBQ3Jtc0tteHJfRm90MTJ4VnlGa1hpTDZxOHhsVHN3ak5VaG9zT1VBMmJQZGg2V3E4Zzk2Zms0dkhjUjFleng3RzVueF8wY0ZlOTRsTG9acDBHczFTZHpLcjZkZks5MTRVckdFRTZkX2YzSTNaX2FDRVpTNDV3dw&q=https%3A%2F%2Fcolab.research.google.com%2Fdrive%2F1PorBmm4bUjY1-dk4f7XKrO7Mx51wbqYU&v=79K9k8OSpIA and getting this problem @sand iris

[sand iris]

did you run everything in order ? I don't use colab much, I really don't know how to help here

[cloud merlin]

yeah

[viscid moss]

Have u tried restarting the environment?

[cloud merlin]

yeah

[glad zealot]

ye ive seen people reporting that same error but no idea why its happening

[viscid moss]

Colab issue

[viscid moss]

Try RVCDisconnected, it's working

[cloud merlin]

ok

[glad zealot]

try kaggle training :D

[viscid moss]

@cloud merlin Kaggle works too

[cloud merlin]

link

[zealous shuttle]

is your kaggle for inference only?

[molten pecan]

https://media.discordapp.net/attachments/1167960398025195590/1185242202012663808/image.png?ex=658ee5d1&is=657c70d1&hm=db1a3bdf17e349d4d40976b9829513a05796edd5da010efdbb5581d56e0ac178&=&format=webp&quality=lossless

...

[glad zealot]

traning, voice changer, cover gen

[viscid moss]

@glad zealot

[zealous shuttle]

i didn't see a training tab there

or maybe i s=just not notice it

[glad zealot]

you select the number on rvc

[zealous shuttle]

so its 1?

[viscid moss]

interesting....

[glad zealot]

1 or 2 yes

[sand iris]

pstt, can that be made into a dropdown ?

[glad zealot]

kaggle no

i wish tho

[sand iris]

ah

[glad zealot]

i doesnt have colab's interface sadly

would be nice if its possible to takeover the website's html to make it work tho

[zealous shuttle]

which one is mostly recomended for training?

[glad zealot]

1

def 1

100% 1

on the gui you can leave it be

[zealous shuttle]

okay then

[serene solstice]

what's an average epoch time for ~18min of data

[elfin cedar]

Hey I have a custom rvc model that a friend made for me, i just downloaded rvc for the first time, how to i go about adding my custom voice and then adding/converting my own voice?

[proper shale]

Are you using W-Okada or some fork of RVC?

[proper shale]

Depends on batch size

[serene solstice]

mine's at 16

just looking for a ballpark number to compare against what I'm getting

[proper shale]

10-20 secs then

[serene solstice]

well damn

[proper shale]

at least on colab

[serene solstice]

getting a minute per locally :/

[proper shale]

what gpu do you have though

[serene solstice]

RX 6700 XT

[proper shale]

...training with AMD? interesting

[serene solstice]

rocm and linux yeah

[proper shale]

i think that's fine, but next time try a lower batch_size

6700XT has only 8GB of VRAM so I'd use 6-8 batch_size only

[serene solstice]

mine has 12

I'm using 11/12 rn with data cached

[proper shale]

oh

[serene solstice]

tried batch 6 earlier and it was like 1m30s

[proper shale]

nvm i was wrong 💀

anyway

[serene solstice]

coulda sworn I got a really low epoch earlier but it could be the sleep deprivation. unfortunately not a lot of data to Google about improving rocm amd training 😭

[sand iris]

low epoch ?

In terms of time?

[serene solstice]

duration

[sand iris]

I swear RVC is bugged as fuck

I've had the same happen to me and like one or two others

Thing works, trains fine

then BAM, slow as fuck

No amount of python venv nuking & reinstall did anything

profiling the code revealed that... time was spent in private pytorch fuctions

[serene solstice]

I made a test model and I swear to god I used the exact same data I'm using now and my epochs were like 6s but I probably accidently only used like 1 file or something idk

[sand iris]

Like two weeks after, I retry and it worked at expected speeds again

[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

How to Make AI Covers using Ilaria RVC

Link: Documentation
Credits: Nick088, Kanav

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How to make an AI cover using Hugging Face 🤗

Link: Rentry
Credits: 👽 Julia (ailen2091)

[sand iris]

How long is the dataset?

[serene solstice]

about 17min

[sand iris]

Yup it was definitely bugged

6s is way too low lol

I've only seen that when it didn't register the dataset for training

[serene solstice]

might be. looks like I was using Applio for that one. have since reverted to rvc web ui

[rare goblet]

Ayo? @serene solstice level 4 !!!

[sand iris]

Somewhat unrelated but you might want to use the latest torch rocm release

If you weren't already

[serene solstice]

I was having issues cause the latest stable is for rocm5.6 but I have rocm5.7 and the rvc repo says rocm5 4.2 is mandatory

[quick jungle]

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

[proper shale]

_ _

[serene solstice]

I'm not sure if 5.4.2 is mandatory or "mandatory"

[quick jungle]

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

[proper shale]

ignore qwq

[chilly peak]

would a bigger batch size make loss_mel go down faster ?

[sand iris]

the rvc repo says that ?

[proper shale]

not necessarily

[sand iris]

I wouldn't trust it too much honestly

there's a lot of misinfo thrown around about ROCm

the "you need a specific version of rocm" being the worst offender

[serene solstice]

"you need to use the following pytorch versions"

[sand iris]

yeah 2.1 is API compatible fortunately

[serene solstice]

then links the rocm5.4.2 pytorch

[sand iris]

at least for what RVC uses

[chilly peak]

i'm at a batch size of 10 for like 300 samples

[serene solstice]

I suppose I can just try a new install with it. worst case it doesn't work and I just revert

I've got the process mostly down now having done it twice

[sand iris]

great lol

[serene solstice]

there an ideal python version for rvc?

[sand iris]

3.9/3.10

[serene solstice]

I've just been installing 3.10.13 with conda

[sand iris]

yeah that'll do

[chilly peak]

is a difference of 1 between loss_disc and loss_gen too much ? or should it really be almost at the same value?

[elfin cedar]

Imma be honest i have a no idea im a noob at this stuff. I wanted to do ai covers so i paid a friend for a model not knowing I didnt have anything to run it through, i have a mac. i downloaded rvc v2 on my dads old windows computer. it worked im just not sure how to actually get my model into it or anything else. Hopefully that cleared soemthing up lol also if you have any suggestions for anything on mac i could use to run my voice models i dont need to be able to train them

[tame mica]

using models is quite easy actually

[proper shale]

oh alrighty!

So, add the .pth to weights and, for the index, make a folder in logs with that model's name.

And, if you own a Mac with the M series of processors (M1, M2 and M3) you can run RVC on them

[elfin cedar]

So i just looked and i have an I9 so does that mean I wont be able to use rvc?

[rare goblet]

Ayo? @elfin cedar level 1 !!!

[sand iris]

It could work but not many have tried

[elfin cedar]

ok

[rich marsh]

is applio works ? colab i mean

[serene solstice]

I think I trained it without waiting for the feature training to finish cause I just reinstalled Applio to try to reproduce the steps. The feature training is less obviously in progress than base rvc lol. it sat there doing jack for like 30s. anyways trained after it actually finished and am getting 1 minute epochs like with base rvc lol

[sand iris]

1min epoch ?

[serene solstice]

ye

[sand iris]

which batch size are you using again ?

[serene solstice]

12

[sand iris]

I do like 24s/epoch for 5mins of dataset on batch size 6 on a 6600xt

[serene solstice]

does that track with 1m/epoch for 17min daya on batch 12 on a 6700xt?

[sand iris]

Oh I thought it was 7mins

[serene solstice]

sounds about so

[sand iris]

Looks good then

mb

[serene solstice]

nice. im trying the updated pytorch rn for the hell of it

had to use a nightly pytorch link I pulled out of the depths of a months old GitHub issue cause stable only supports rocm5.6 rn

ehhh maybe faster by like 5 seconds lol

oh maybe more. epoch 2 at 56s with batch size 12 huh

[cloud merlin]

help

[turbid badger]

does anyone have an alternative to ai cover gen that can be run externally / online since colab is blocking it rn?

[brittle wing]

does anyone have a tutorial for creating a model?

[elfin cedar]

would u be able to help me in a private message i could send you pictures of my screen?

[brittle wing]

Hey, I'm trying to get RVC running locally on my mac. Looking around, it seems like I should be building Mangio or Applio from source

Is there a one click installer or docker image? Am I even on the right path? Couldn't find too much info in the guides

[sand iris]

applio supposedly has an install script

I think mangio has a binary release for m1/m2/m3 macs

...But there's always mainline which also has a single install script (though it has to be fixed first)

[brittle wing]

Ayayay

What's your rec? I can probably manage any of the above

[brittle wing]

Going to try Mangio, just need to raw dog this script into my terminal 🤞

[sand iris]

mangio hasn't been updated in a while
applio has proven to be highly unstable
mainline's script is borked (though the fix is like a one-liner)

[brittle wing]

Thanks

[rare goblet]

Ayo? @brittle wing level 2 !!!

[turbid badger]

? :((

[thick trout]

hey uh anyone here to help with NameError: name 'experiment_name' is not defined

folder name does not contain any special letters

and all the files are wav i dont get it

no support here?

[fossil glen]

[white bough]

If you're using the colab, you probably haven't run one of the first cell, that sets the parameters

Anyone using kaggle here knows how to download files from the Output? I can do that with small files, but with bigger ones, I click the Download button and nothing happens

[elfin cedar]

can anybody help me 1 on 1 on how to use my voice model on rvc v2, ive never used it and i dont know any of these coding words, im a noob

[thick trout]

you were right. I forgot 1 cell

[brittle wing]

how much of an performance improvement can i expect from running it on linux compared to windows (7900 XTX)?

not even sure if it supports rocm 😅

[thick trout]

my dataset is literally 4 isolated vocal wavs, is that normal or should i have added more songs? This is my absolute first model so.

the quality of the isolations are not perfect too so should i expect something crappy

[chilly peak]

is it normal that loss_disc/gen/mel don't really go down but stay relatively constant ?

[snow hazel]

How do I update appilo?

[chilly peak]

or is this just a matter of waiting to get a lot more epochs, because i'm only at like 90 ish

[fiery torrent]

why when i try to train feature index on applio i just says "Generating training file...
Generating adding file..." and doesnt do anything

[snow hazel]

I need help

[mossy ferry]

Is there any way to give a song but have the ai sing different lyrics to it? I've seen it all over but i wonder if i need to put myself into the ai or not

[thick trout]

guys there is no .pth file after model creation

[sand iris]

Yo don't post instrumental in here

[supple ibex]

its copyright free

[sand iris]

is it ?

[supple ibex]

wait

i need to check

ah sorry it isnt

my mistake

[sand iris]

I nuked it because we'd rather just not take any risks

ya

[supple ibex]

because the artist has some older songs which he recorded himself without a label

let me make another cover without instrumentals

[molten pecan]

just reinstall it

[sand iris]

https://cdn.discordapp.com/emojis/739053025125204041.gif?size=48&name=kkkkek&quality=lossless

[hollow bluff]

hi please easy gui link

[sly temple]

Anyone know how to get the Easy GUI to work locally? For some reason it says that the Utils.py can't import a file

[vague oriole]

I'm using mangio rvc v23.7.0, i know that a model's .pth file goes in the weights folder, but im not sure where to put the .index folder
any help?

[acoustic carbon]

if you train, it will be under the logs folder

[vague oriole]

im not training

[acoustic carbon]

or just link the path of where it is

[vague oriole]

oh

[acoustic carbon]

-rvc

[azure marsh]

Suggestion for @hollow bluff

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

[acoustic carbon]

what have you used to train? it should be under the weights folder

[violet heron]

Index goes in logs folder

[acoustic carbon]

told him that

[sly temple]

Does anyone else have a problem opening up EasyGUI locally? I'd rather use Google Colab but it keeps cutting my runtime

[acoustic carbon]

what exactly is the problem? any error codes, why does easygui not start?

[violet heron]

EasyGUI on colab doesn’t work anymore

Do you want to make covers or make models?

[sly temple]

I want to make covers

[rare goblet]

Ayo? @sly temple level 1 !!!

[violet heron]

-colab Ilaria RVC, CREATE A HF ACCOUNT THEN DUPLICATE THE SPACE OR IT WONT WORK

[azure marsh]

Suggestion for @sly temple

All working colabs / spaces ☁️

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

Credits to Faze Masta and Antasma for compiling these links.

[sly temple]

Also Im talking about running EasyGUI locally using the downloaded Mangio.

[violet heron]

Oh

[acoustic carbon]

well what is the error code? have you installed it correctly?

do you have python installed?

[sly temple]

I do have python installed WSL

[violet heron]

Does normal Mangio work tho?

If it does why don’t you just use that?

[sly temple]

ImportError: cannot import name 'CSVutil' from 'my_utils' (A:\Mangio\Mangio-RVC-v23.7.0\my_utils.py)

[sly temple]

For some reason it just loads and nothing happens, I receive a bunch of errors in the command terminal when I restart the page

[violet heron]

Which file are you opening?

[acoustic carbon]

get the my_utils.py from the mangio kalo tweaks

looks like you havent installed mangio correctly

[sly temple]

I'll try that. Thanks

[sly temple]

Still running into issues.

[snow hazel]

I tried. But it still said I need to update it.

[acoustic carbon]

maybe try moving it onto your c drive

[molten pecan]

https://github.com/IAHispano/Applio-RVC-Fork/archive/refs/heads/main.zip

did you try from there?

[acoustic carbon]

or open cmd in mangio folder and put runtime\python.exe -m pip install fastapi==0.88.0
and then runtime/python.exe, since i think you really havent gotten everything necessary installed

[sly temple]

I appreciate your help, that worked. How did you figure that installing the fastapi would work?

[acoustic carbon]

looked like some dependencies were missing and fastapi includes that

[vapid gust]

De

[rapid grail]

hellooo

[rare goblet]

Ayo? @rapid grail level 8 !!!

[snow hazel]

So what do I press 1 or two for the installation?

[feral saffron]

Hey can someone help me? Does all-feature-done mean that feature extraction is done?

@proper shale sorry for pinging

Can you help me with this?

Never mind I figured it out

Thank you tho

[molten pecan]

@brittle wing

[acoustic carbon]

type 1 for nvidia and 3 for amd

https://docs.aihub.wtf/vocal-cloning-guide/applio

read more here

[keen stag]

anyone know a good gui to train locally?

[molten pecan]

yas

[molten pecan]

what's a GUI?

[acoustic carbon]

-rvc

[azure marsh]

Suggestion for @keen stag

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

[acoustic carbon]

stock rvc or mangio

[keen stag]

how are vocal cloning and creating a model different?

[acoustic carbon]

same thing

you need to train a model first

[proud elbow]

vocal cloning is inference using a voice model

[acoustic carbon]

a model is the thing that clones the voice

inteference is basically making a ai cover or something like that, using the model

[brittle wing]

these sttins are okay for aa 6600 xt?

[acoustic carbon]

change the extra to 16384 if you can

[brittle wing]

i cant

[acoustic carbon]

not even if you stop it first?

shouldnt be like that i think

the other settings look good though, might need to adjust tune and s.thresh if its not good

[brittle wing]

i cant use it in real time cause y get horrible results

@acoustic carbon wich i need download if i have an radeon?

[acoustic carbon]

MMVCServerSIO_win_onnxdirectML-cuda

[brittle wing]

i got it

[acoustic carbon]

i thought you installed okada already

[brittle wing]

Yes, I have it installed, but it was to see if I didn't have the wrong version

i cant hear my voice when i use it, it get too distorsionata

[acoustic carbon]

have you installed vb cable and setup the right inputs/outputs

[rare goblet]

Ayo? @acoustic carbon level 5 !!!

[brittle wing]

yep

[acoustic carbon]

just read this 🗿 https://rentry.co/VoiceChangerGuide

[brittle wing]

okay

thx

now i get better results, but input lag

[rare goblet]

Ayo? @brittle wing level 2 !!!

[snow hazel]

Which one though. That’s what I’m asking.

[molten pecan]

1

[snow hazel]

It says I don't have python installed but I do

[brittle wing]

my mangio has this error every time can someone help? : Expecting value: line 1 column 1 (char 0)

[fossil glen]

finally @quick jungle work

[quick jungle]

Watchu want? :3

[fossil glen]

@quick jungle

[quick jungle]

Watchu want? :3

[fossil glen]

qwq

QWQ_Chan

[snow hazel]

What do I do?

[unkempt forum]

pls help, I cannot find PTH file after complete model train, just found index file only 😦

[unkempt forum]

thank you, I trained on local, not Colab, I've found index file but cannot find PTH

how to get it on local

[fossil glen]

its on rvc/assest/weights/yourmodel.pth

[brittle wing]

my mangio has this error every time can someone help? : Expecting value: line 1 column 1 (char 0). Im trying for 3 hours pls

[unkempt forum]

I can't even found assest folder inside RVC folder

[rare goblet]

Ayo? @unkempt forum level 2 !!!

[fossil glen]

find weights with ctrl +f

[snow hazel]

Help

Oh. I just read it. Do I really need to install python on microsoft store?

[unkempt forum]

There's no match result. I've trained model with one-click function on Gradio but only index file generated after completed notification, I think the model file PTH may not be generated, right?

[snow hazel]

Yes

[viscid moss]

Install Mangio

[snow hazel]

Link please?

[viscid moss]

https://youtu.be/ixB9oalT3cQ?si=fs7e-C_Q0aON2kyw

[snow hazel]

Does appilo doesn't work anymore?

[viscid moss]

Applio works but installation is a problem

[snow hazel]

Oh.

[brittle wing]

pls help me with this line 1 column 1 (char 0)

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

😦

[snow hazel]

I keep getting this

[copper pebble]

check your internet
connection/make sure the remote file exists and verify that your dependencies are up to date

check the link too

[final tulip]

for some reason my vocal results end of being like double the length of the original input audio and its super slowed down and deep pitch even when using base settings which were working fine before

i fixed it nvm

[slow hill]

Link

Of rvc V2

Plz

Song convert

[azure obsidian]

bruh

[finite galleon]

Lol

[azure obsidian]

is rvc v2 better

idk

[finite galleon]

I was about to reply

[azure obsidian]

i had rvc v2 long time but it got deleted

😭

[finite galleon]

What?

[azure obsidian]

i had rvc v2 on my google drive but i kinda deleted it

[finite galleon]

What do you mean brotha

[finite galleon]

Ohh your models?

[azure obsidian]

yes and google colab too

i couldnt find all of them

[finite galleon]

If it hasnt been 30 days it should be in your trash bin

[azure obsidian]

i will keep using ilaria rvc it looks good

[finite galleon]

Yeah ilaria is user friendly

[azure obsidian]

wait does rvc v2 models works?

[rare goblet]

Ayo? @azure obsidian level 2 !!!

[azure obsidian]

i remembered some of my models but idk if they work in ilaria

[finite galleon]

Yesh

Yeah*

[azure obsidian]

oh ty i will download more models

[slate pawn]

can anyone help me? the index file is not working on the RVC

[sand iris]

which version are you on ?

I'm trying to see where something is going wrong, but the line numbers don't really match with latest

[proper shale]

Yes! Sorry for the late response btw

[slate pawn]

This is version 1.5.3.10 for the RVC

index hasn't been working for the models I train, but downloaded models index files works great

[proper shale]

Try going to 1.5.3.15

[slate pawn]

yup I'm currently downloading it

[brittle wing]

yo wsp

i need somehelp with the delay

yeah its work

but the delay is mad awful

how can i fix dis

[proper shale]

first off, what GPU do you have

[brittle wing]

ughhhhhhhhhh yeah alright it gonna seems crazy what im about to say

i use cpu 💀

i mean i cannot change the gpu its automatically set to cpu

[proper shale]

oh so integrated?

[loud summit]

My model keeps collapsing all the time.

[proper shale]

well I'm just gonna say that it ain't gonna work that great if that's the case

[brittle wing]

what is a intergrated

[rare goblet]

Ayo? @brittle wing level 1 !!!

[proper shale]

you'd need an NVIDIA GPU, at least 10+ series

[proper shale]

inside the processor, basically

[proper shale]

if the quality gets better it shouldn't be a problem

anyway, idk why this happens, I'd say just trim silences to 0.038 and try again

[brittle wing]

does gtx 1660 work 6g

[proper shale]

yeah maybe

[brittle wing]

ight so rn how do fix it

just temporary solution

[loud summit]

Uh, okay, that's not a problem for me.

[proper shale]

select the GPU

[brittle wing]

so should i delete everything and download it again?

[brittle wing]

the setting is set as cpu

i cant change it 1

[proper shale]

why?

isn't there a gpu dropdown

[brittle wing]

tbh idk 2

i wish

[proper shale]

huh...you have a 1660 right?

[brittle wing]

ah

nah

ima buy it nah not now

[proper shale]

oh

try using dio as ur f0

[brittle wing]

alright

[proper shale]

ig that's the best config for u rn

[brittle wing]

ok let me see

[rare goblet]

Ayo? @brittle wing level 2 !!!

[brittle wing]

oh yh the delay less worse but the voice is horrible

sound like they are crying 4 help

[slate pawn]

index is still not working

on updated RVC

what's funny is that all index files work but the files I trained myself so I probably did something wrong but I don't know what it is lol

everything went smooth in training but the index files

[chilly peak]

does loss_mel not going down mean that the datasets are too diverse? i.e. there are too many pitch differences in a lot of the samples

[chilly peak]

or does loss_mel not necessarily need to be near 0 to have a good model ?

[wooden ruin]

i probably sound stupid but what do i install do use rvc

[supple ibex]

dataset vs output, why does it sound so bad (400 epochs, 30 minute dataset, 48k sample rate)

ignore the first 5 seconds of the audio

[craggy cobalt]

How does RVC work? is it just a program that you feed models with audio datasets to and it layers it onto an audio file you give it to type of thing? what does it stand for? is it new?

[supple ibex]

basically yeah, it trains based on audio clips you give it and then when you give it for example a song, it puts the trained voice on the song

[craggy cobalt]

cool, what does it stand for

?

[supple ibex]

retrieval based voice conversion

[craggy cobalt]

is it new

??

[supple ibex]

it's been out for ~6-10 months

[craggy cobalt]

it's pretty new then

[rare goblet]

Ayo? @craggy cobalt level 1 !!!

[supple ibex]

it's what they use to make those ai covers

like spongebob singing ballin

[sand iris]

Retrieval-based Voice Conversion

[supple ibex]

also can you help me

[craggy cobalt]

is RVC a method or just the program itself

[sand iris]

The programs label themselves as RVC, but yeah, it's the underlying method

RVC was born like what, around 6monthd ago?

[craggy cobalt]

who developed it? was it like devloped here in this server?

[sand iris]

I don't train models, not sure how I could help much, good luck though!

[sand iris]

I think it was a team of students in Asia primarily, but there's a few contributors to the project here. No one from the "core" team though

[tough slate]

yo guys im trying to use the voice changer on roblox but it doesnt connect and i sound like myself,how do i fix that?

[craggy cobalt]

now that i think about it I remember seeing some stuff on github related to rvc in chinese, very interesting

[static widget]

So i’m using Applio RVC on HuggingFace, and whenever I use the audio separater, the instrumental comes out very crackly and weird (on high volume/bass parts maybe?) and I don’t know what the issue is. The reference audio is completely normal. Help?

[proper shale]

Just use MVSep or Ilaria UVR for separating stuff. Applio's separation has always been meh for me

[versed gate]

[Voice Changer] VoiceChangerManager initializing...
[Voice Changer] Web Server Launch Exception, Expecting value: line 1 column 1 (char 0)
[VCClient] wait web server...10 http://127.0.0.1:18888/
[VCClient] wait web server...20 http://127.0.0.1:18888/

why isnt it workin

it worked the firs time

[proper shale]

do you have a firewall?

or any antivirus active

[versed gate]

but it worked before, how do i check anyway?

[rare goblet]

Ayo? @versed gate level 1 !!!

[versed gate]

cool stuff

[proper shale]

wait, then try this:

delete stored_setting.json, restart W-Okada

[versed gate]

wheres stored_setting.json

[proper shale]

in the W-Okada folder

[versed gate]

you sure

i should delete it?

deleted it

ill try again

works

what was the issue?

[proper shale]

not sure what it is, but thats how to fix it lol

[versed gate]

another issue that i got is

[copper pebble]

[versed gate]

oh nvm

figured it out

time to try text to speech xD

[copper pebble]

11labs or ilaria rvc pretty good

11labs has a max of character per month

ilaria rvc is unlimited

[versed gate]

https://github.com/IAHispano/Applio-RVC-Fork

isnt that good?

[copper pebble]

yh u coudl also use it

[versed gate]

its what im gonna try

seen it on yt

https://www.youtube.com/watch?v=jh83ViDUvJs

[versed gate]

holy

[sand iris]

lolz

[subtle verge]

[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
How to fix

[versed gate]

what should i do if i clicked and it got stucc?

[rare goblet]

Ayo? @versed gate level 2 !!!

[versed gate]

ayo

[versed gate]

clic enter

[versed gate]

progress xd

[snow hazel]

How do I check if my dependencies are up to date?

[rare goblet]

Ayo? @snow hazel level 6 !!!

[scarlet relic]

ok so when im just in the program the audio is just fine, but when i try to send a voice message with the VB Cable method (through vencord pc) the audio is a lot more robotic and weird. it worked more or less fine before but idk whats hapenning now

[proper shale]

try VAC by Muzychenko, VB cable has bugs for some people

[brittle wing]

someone help me with this problem: Expecting value: line 1 column 1 (char 0)

[versed gate]

delete stored_setting.json, restart W-Okada

[young halo]

How can I post my model here?

[molten pecan]

-huggingface

[azure marsh]

Suggestions for @young halo

huggingface.co upload Guides

English guide

Link: https://rentry.org/fdg_guide_new
Credits: FDG

Spanish Guide

Link: https://rentry.org/AI_Hub_guide_esp
Credits: Julia (ailen2091)

Hindi guide

Link: https://rentry.org/AI_hub_guide_hi
Credits: Enes

Turkish guide

Link: https://rentry.org/AI_hub_guide_tr
Credits: Enes

[rapid rose]

so uh it says "no valid nvidia gpu found" is the RTX 2060 not working with RVC?

[molten pecan]

uhh it should

[rapid rose]

i also get a bunch of errors

[molten pecan]

maybe uninstall RVC, reboot your puter, and install it again

[molten pecan]

let me guess, you using Applio?

[rapid rose]

??

i just followed the RVC read me on github

[copper pebble]

3060 is reccomended

so

[rapid rose]

-._-.

[rare goblet]

Ayo? @rapid rose level 1 !!!

[molten pecan]

can you send the link?

[rapid rose]

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

[rapid rose]

im able to use sd webui on my 2060, idk how heavy rvc is.

[versed gate]

im trying to listen to the file and i cant why?

[rapid rose]

so uh

[molten pecan]

that's main line RVC. The OG RVC and not a fork

that's good

it's easier to install compared to Applio, the other popular fork

[rapid rose]

oh

[molten pecan]

@silver anvil wait why meat ball emote 😭

meatball

meat ball

[silver anvil]

oh

that funny

[molten pecan]

WHAT

dang

[versed gate]

bruh

[rapid rose]

so uh.. which fork should i get

[molten pecan]

😭

[versed gate]

voice sounds so creepy

XD

how do i get a more realistic one

[molten pecan]

Applio: A fork of Mangio RVC, developed by IA Hispano. This version of RVC is mainly chosen due to its user-friendly aesthetic and additional features, like TTS, voice model download through links, music mixing options, and others.

Mangio: is a [fork] of [RVC], made by Mangio621. Its main feature is hybrid training (the ability to use multiple extraction algorithms for training a single model) and user interface for CLI (a way to run an app from just a terminal, without a GUI).
Mangio is also liked for being easier to install compared to Applio.

[rapid rose]

I see

[molten pecan]

you are using the OG RVC. I recommend staying with that one

but if there is no way to fix the errors, go for Mangio

[molten pecan]

you know what a fork is?

[rapid rose]

yes

[molten pecan]

great

[versed gate]

the TTS on applio is just a wav file, is there any way to use it as a mic?

[molten pecan]

you mean like, you want to talk through a mic and make Applio convert into TTS in real-time?

[versed gate]

yes

well no

write TTS stuff > RVC > mic

basically

type "hi" on TTS > RVC convert to anime girl voice and then make it sound through the mic

i mean Applio may be a AIO but how do you set that up?

[rare goblet]

Ayo? @versed gate level 3 !!!

[molten pecan]

ohh yeah I see. You are going to need a virtual cable for that

but why do you want that though 😭

[versed gate]

i know but is it possible if yes how?

i dunno want to try it

xd

[molten pecan]

Windows
https://vac.muzychenko.net/en/download.htm

Mac
https://vb-audio.com/Cable/index.htm

find a tutorial on youtube

[versed gate]

yeah i already got that but im asking

how do i make applio work

do i use the wav file on rvc or ?

[molten pecan]

in case you didn't know, you can do inference in real-time through mic input without a TTS

and they won't know your real voice

[versed gate]

living with family tuff

xd

[molten pecan]

then no idea

I think you should use another tool for that. In Applio takes an extra step

[rapid rose]

I'm gonna try mangio

[molten pecan]

a bit too long

[molten pecan]

good luck!

[versed gate]

any suggestions?

[molten pecan]

idk just search "tts to anime voice app" on google

[rapid rose]

im installing mangio and applio. so if all hell fails, ill try applio as a last resort.

i believe i setup mangio proprely. am bout to try it.

[rare goblet]

Ayo? @rapid rose level 2 !!!

[rapid rose]

where do i put my voice models

[versed gate]

make a voice folder O.o

[rapid rose]

o

[versed gate]

i have a question

why does my voice sound the same although i change models?

basically it isnt working at all

it worked before for a second but couldnt get it to work again

same exact robot voice on every model

[rapid rose]

mangio was unsuccessfull

somehow applio while being annoying to setup a bit. was able to get it running

i was actually able to start a conversion

[wispy geode]

@molten pecan when I tried to make hugging face account it shows an error and I couldn't make it!!
(Sorry to bother you)

[rapid rose]

tho it still says no supported gpu found and tries to use my cpu instead

[molten pecan]

what error?

[rapid rose]

like, mangio, og rvc and applio all says no supported gpu

[molten pecan]

what GPU do you have?

[rapid rose]

RTX 2060

[molten pecan]

maybe it doesn't support that one

[rapid rose]

it should

[molten pecan]

@proper shale thoughts?

[rapid rose]

RTX 2060 has cuda and stuff

im able to use it for stable diffusion web ui

[molten pecan]

when did that happen?

when you clicked the confirmation link?

or when you logged in?

[wispy geode]

When I tried to create a account

[wispy geode]

Yes when I tried to sign up I did this twice but still it's same

[molten pecan]

are you using a VPN?

are you doing it on mobile?

[wispy geode]

No

[wispy geode]

Yes

[molten pecan]

try doing it on a computer then

or in another web browser

like Brave or Firefox

[wispy geode]

Ok then I'll try this on PC

[rapid rose]

using my cpu, it took about a minute and a half to convert an audio

am surprised it didn't take longer than that

[molten pecan]

how long was the audio?

hola Light

[rapid rose]

45 seconds

[molten pecan]

queeeeeeee en serio?

[rapid rose]

well all 3 of them says "no supported nvidia gpu found"

including applio when i launch it

am not training

[molten pecan]

uses*

te faltó la S al final en este contexto

doesn't*

:b

Applio es un objeto (it). Doesn't se usa cuando el objeto es una persona o cosa

Don't se usa en I, we you o they

I don't use Applio
Julia doesn't use Applio

entiendes?

xd

[rapid rose]

i don't speak spanish

[molten pecan]

@rapid rose

@viscid moss Le puedes ayudar?

Applio no le detecta la GPU

[molten pecan]

Es compatible con RVC?

[rapid rose]

I don't speak spanish ;w;

[rare goblet]

Ayo? @rapid rose level 3 !!!

[molten pecan]

dw I'm describing the problem to a Helper

[rapid rose]

o

RVC also gives me that same error tho

[young halo]

Can someone help me trying this model?
https://huggingface.co/Ryujin88/AVA_MAX_HEAVEN_AND_HELL/resolve/main/heavenxhellava.zip

I wanna know if this just happens to me with the errors

[lavish meteor]

What is best collab now?

[obtuse herald]

guys please help me
batch convert:
Traceback (most recent call last):
File "D:\Folders\aishit\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\infer-web.py", line 316, in vc_single
audio_opt = vc.pipeline(
File "D:\Folders\aishit\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\vc_infer_pipeline.py", line 539, in pipeline
sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
RuntimeError: CUDA error: the launch timed out and was terminated
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

[rare goblet]

Ayo? @obtuse herald level 2 !!!

[molten pecan]

For training or for inference?

[lavish meteor]

Training a voice first.

[viscid moss]

busy rn sorry

[molten pecan]

Training
https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=Sb5fzhzEXK8X

Inference
https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp=sharing

[molten pecan]

Suerte!

[lavish meteor]

Thank you so much.

[rare goblet]

Ayo? @lavish meteor level 1 !!!

[molten pecan]

If you need guides lmk

[radiant pendant]

Does anyone know, is there some kind of epoch calculator? It would be good for model training

[obtuse herald]

can you help me?

[obtuse herald]

...

[rapid rose]

that can't be right

[obtuse herald]

its was working till yesterday

thats the problem

[rapid rose]

I'm able to use my GPU for stable diffusion web ui image generating

[small umbra]

THIS PERSON IS LYING!

[radiant pendant]

?

[rare goblet]

Ayo? @radiant pendant level 1 !!!

[rapid rose]

No the hell i'm not. I can prove it >:(

[small umbra]

THIS PERSON IS LYING!

[rapid rose]

Bet

let me launch it

[steady ruin]

is this normal?

i mean for the percentage to be 0 even though it's halfway there

[rapid rose]

I get that on my thing aswell

[steady ruin]

what I've just posted?

[rapid rose]

yea

[steady ruin]

ok

and it works fine?

[rapid rose]

oh wait you're training a model?

[steady ruin]

yeah

[obtuse herald]

yep its okay

[steady ruin]

ok

[rapid rose]

ohh, idk then. i get the same issue when infering but still works fine. idk about training tho.

[steady ruin]

thanks

[obtuse herald]

that means it started

(new epoch)

[steady ruin]

oh i see

previously the percentage went up

so i was worried something went wrong

[obtuse herald]

nope
just finished a model and it was exactly like this

[lavish meteor]

It would be good actually.

[modern surge]

I have this error, I don't know what it is, but it slows down the time of making a cover a lot

[steady ruin]

does your model sound good?

[steady ruin]

yours i mean

[obtuse herald]

eh,50/50

[steady ruin]

was that because the 0% thing

[modern surge]

Someone else happens?

[obtuse herald]

my vocals were not clear,thats the only reason

[steady ruin]

oh, you're making a model of yourself?

[obtuse herald]

yep

[molten pecan]

Training guide
https://docs.aihub.wtf/guide-to-create-a-model/model-training-colab

Inference guide
https://docs.google.com/document/d/e/2PACX-1vThk7Qo7yCWNVbxOmahl2R8_Jgi6TFuMBUIi-PWre_HIN0lFTq-dr37Rh5iJlGgYb_vFapXMHt2W8Kp/pub

My list of guides and apps:
https://rentry.org/aihubstuff

[steady ruin]

aha

[modern surge]

I repeat

vith error: libcufft .so. 10:

[molten pecan]

Is that Colab?

[rapid rose]

hah xD

[sand iris]

Yup

You want a fix ?

WAIT

.10 ?

What

Which notebook even is that ?

[obtuse herald]

congrats!you won vs a discord fun bot!

[proven hill]

i can be wrong but this is not rvc

jk lol, better ask on #🏙│ai-images

[rapid rose]

I wasn't asking anything lol. i was proving i can ai generate images using my gpu cuz the bot said am lying xD

[brittle wing]

smbdy knows what to do?

[sand iris]

Oh you poor soul

This is what we call an "applio moment" here

Really sorry for saying this, but you might consider using the "Original" RVC, unless there's a feature Applio has you absolutely need (Can't think of anything rn)

[rapid rose]

i was able to get applio working

with my cpu atleast

[sand iris]

[brittle wing]

Ok thanks im trying that

[sand iris]

Do you need a link?

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

[brittle wing]

Thanks 🙂

[sweet cradle]

Is there a way to use premade models on Mangio? I have the model's path and index file in the Logs folder, but the program only sees the index file.

[slim sky]

is there any sort of lite version of applio where its just the sample audio in, sample audio out? my beefy PC is shared and I'd like to do some voice stuff w my models i made on my underpowered personal PC

[copper pebble]

pth needs to be in weights's folder

[slim sky]

actually id like to use the RVC v2 huggingface space but im not sure how to get my models in it?

[calm shadow]

i cant send photo

[small umbra]

THIS PERSON IS LYING!

[molten pecan]

https://rentry.org/ilarvc_inf_guide use my guide :b

@small umbra

[calm shadow]

ا

wait web server...

@molten pecan

[molten pecan]

What?

[calm shadow]

how i can add model

[rare goblet]

Ayo? @calm shadow level 1 !!!

[calm shadow]

@molten pecan

[slim sky]

thank you @molten pecan !!! but where's the three dots to duplicate the space 💔 i only see a hamburger menu

[rare goblet]

Ayo? @slim sky level 1 !!!

[slim sky]

wait now i dont even see the hamburger menu

OH!! found it sorry

uh oh

all i did was duplicate the space was i not supposed to rename it something silly...

[molten pecan]

Reload the page, dw

[molten pecan]

What's dat?

Applio?

[sand iris]

voice-changer

[molten pecan]

Ah

[slim sky]

no more error but it still says building! do i just wait? :o

[molten pecan]

I don't know how to use the Voice Changer then

[molten pecan]

Yeah of course

Like I said in the guide, wait a few minutes for it to load

[slim sky]

okay!!! haha i didnt see that im silly thank you so much!!! <3

[molten pecan]

yw!