✨│ai-help

[keen vault]

how can I fix that problem?

[zinc sleet]

okay ty

[gray crow]

hey, how was Itaila pre-trained model actually trained?
Can sombedy please help me out how to train my own pre-trained?

[nova trench]

Hello! Is there any way to make vocal extractions sound smoother in UVR? I'm using UVR 6 Karaoke under VR Architecture, but it sounds a robotic / muffled in some parts

[hot lagoon]

witch one is good for duets with female and male singing?

[proven hill]

karaoke 6 is very heavy

i personally would suggest usuing MDXC InstVoc

[proven hill]

do you have an A100?

[gray crow]

yes

[proven hill]

then you can train your own pretrain on there

[gray crow]

First of all, again thank you for replying. As always 🙂
Could you share more details on the training process itself?

[proven hill]

you need countless of hours of dataset, we talking about hundreds

for the rest you just train it like a normal model, then you extrapolate the G and D tensors

[gray crow]

I dont want to create a model of my voice, so if im correct i can create a multi speaker dataset correct?
For the training process.. well okay, thats easy 😄

[proven hill]

wait you want to create a model or a pretrain?

[gray crow]

pretrain! 🙂

[proven hill]

then yes multi speaker dataset is fine

it needs to be cleaned, raw, dirty, different types of audios, so when trained it gets trained on solid data

[gray crow]

I need to train on dirty audios also? But the rest of the training remains the same as with a normal model?

[proven hill]

yes and yes

but again we are talking at least 100 hours for a very good pretrain

[gray crow]

Absolutely.
Ilaria, THANK YOU!! ❤️

[hot lagoon]

Is there a way to separate 2 singers like a duet?

[hybrid narwhal]

huys is there a voice changer avilable ?

[rare goblet]

Ayo? @hybrid narwhal level 1 !!!

[potent root]

when making a cover, how much should i transpose in the voice such that it still sounds good with the original instrumental key

[finite galleon]

A whole octave (12 semitones)

[fading drum]

finding a higher "quality" of your song, can give it more chances to fill in the gaps, but if that doesnt work, then theres nothing that you can do besides wait for a new model, perhaps use lalal.ai in the meantime to see if thats better (you get 5 mins free, just make throwaway gmail accounts and you get unlimited free trials)

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[potent root]

that makes it sound too chipmunky

[brittle wing]

AI voice changer echoing everything that plays in the background, even when other people speak, it plays back their voice in the AI. This applies for everything that contains audio, is there any way how to fix this?

[rare goblet]

Ayo? @brittle wing level 2 !!!

[white bough]

If you keep the original instrumental, you can only transpose the voice by multiples of one octave: 0 octave (0 semitone), 1 octave (12 semitones), -1 octave (-12 semitones) etc.
But then, of course, it might not be a natural key for the singer. So you might have to change the key of the instrumental as well

[rugged crown]

Rvc

Help rvc

Omg how do I get to easy gui

[rapid wharf]

so im having some trouble opening the rvc. i followed the docs guide and double clicked the "go-web" file. it opened the cmd promt and asked me to "press any key to continue", so i did. after waiting a bit, nothing popped up in my browser for the rvc like it was supposed to. so idk maybe im just stupid-

[proud elbow]

python not found error?

[low shard]

Did you follow our docs ?

[rapid wharf]

it just says that the system couldn't find the path specified

[stoic basin]

hey does anyone use the okada rvc

[agile flume]

theres a bug on the weights.gg website

[rare goblet]

Ayo? @agile flume level 1 !!!

[agile flume]

wut

[rapid wharf]

after doing some research i might've found the problem. i apparently downloaded a compressed version of the rvc and now im downloading the full package so it should work by the time its done downloading

[left lava]

-audio

[azure marsh]

📚 Audio Guides

• Perfecting Audio Isolation on Low-End Rigs: A Practical Guide, by Litsa The Dancer and Faze Masta Google Docs
• Gathering and Isolating Audio, by SCRFilms Google Docs
• Creating Datasets for RVC using iZotope RX, by LUSBERT Rentry
• Vocal Mixing Tutorial, by Roomie YouTube

🛠️ Tools

Audio Separation/Isolation

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

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

[slow aspen]

Hey which khz rate should I use for training?

[hardy hollow]

hey, where do i find the rvc thingy on colab? i've used it before but i removed the bookmark and i can't find it anymore

i googled, checked my history

[pastel oak]

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

Ill link this one for now, on the left side you can see the others if you used any different

[hardy hollow]

i didn't mean for training lol

i already have a model

eh im just gonna use ilaria

[proud elbow]

seems ok to go 40k, but u can try low pass 16 khz and go 32k if it might help reducing glitchy sibilances

[slow aspen]

Then it will sound better? if so how to do it?

[pastel oak]

How do you low pass btw? And would it technically make a difference if i dont lowpass my 40k dataset and train on 32k right away?

[proud elbow]

I said it might be 2nd choice, but as the guide says, use the sample rate according to the audio cutout, in the case above it's truly 20 khz cutout so it's fine to go 40k

[robust sun]

Any one suggest me about cloud gpu?

[proud elbow]

free colab T4 GPU

[slow aspen]

Well now I don't know what to believe. Because I just checked the same file in Spek and it came up with this:

(the previous result was from the applio analyser)

[robust sun]

Which colab gpu is best for realtime?

[brittle wing]

mic is too quiet

[hardy hollow]

i found the one that i used but i have another problem

[rare goblet]

Ayo? @hardy hollow level 1 !!!

[brittle wing]

my microphone is quiet when using rvc

please help

[proud elbow]

u didnt modify the audio and have checked also in audacity?

[brittle wing]

also why is my gpu blazing when using the rvc

[proud elbow]

or probably processed through UVR?

[slow aspen]

Yes UVR

[robust sun]

Thanks

[hardy hollow]

take a pan and start frying

[brittle wing]

is this normal

[slow aspen]

But I got two different results from the same file. Which one is the real one?

[rare goblet]

Ayo? @slow aspen level 3 !!!

[brittle wing]

That explains why when I had a 40k audio cutout and trained it on 32k it collapsed 💀

My dataset had noise Litsa told me to denoise & retrain on 32k and all I got was a failure

[proud elbow]

might be harshly cut to 32k instead of applied low pass first

[brittle wing]

Explain

I used Uvr denoise

[proud elbow]

did UVR cut out below 20 khz?

[rare goblet]

Ayo? @proud elbow level 31 !!!

[brittle wing]

Wtf

I have no idea

[proud elbow]

check the results in spek or audacity spectrogram

[brittle wing]

The samples are 44100hz with & without denoising

[proud elbow]

not the audio format sample rate, but the cutout observed in spectrogram

[slow aspen]

@proud elbow Should I trust Spek or Applio analyser?

[brittle wing]

I don't use a computer

[rapid wharf]

nvm its still not working 😭

[proud elbow]

go find similar app for mobile

[brittle wing]

Why don't find one for me?

You

What are the keywords for it

[brittle wing]

I don't need it

[rapid wharf]

has anyone found a fix for this yet cause all i want to do is make a dr. don't ai cover 😭

[rare goblet]

Ayo? @rapid wharf level 2 !!!

[brittle wing]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[rapid wharf]

i took a ss of it

so i downloaded python 3.11 and when i try opening go-web it just keeps coming up with this

[pastel oak]

Are you aware that you downloaded the code which requires everything to be installed manually instead of using one of the binary releases that is pre-installed?

Just asking in case you did not mean to do everything manually

[rapid wharf]

are you talking about "Retrival-based-Voice-Conversion-WebUI-main"?

[pastel oak]

Yes, that is the code itself you downloaded. It requires manual installation, which is also listed on the github

Or you can directly install the package that comes with everything installed, which I would advise you to do if you dont exactly know how to manually install it

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases/tag/updated1006v2

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[brittle wing]

Is it possible to stop the cell and export the model.at 100 epochs?

[bronze maple]

Question for the Pros: When I enter the path of my dataset on RVC, then it processes the data and cuts it, so I go look for this folder Applio/Logs/ModelTest/1-16-wavs and when I listen the quality is disgusting, it's is normal ?

[ruby rain]

What good quality voice recording program do you use for Okada Realtime voice record of vocal?

[rare goblet]

Ayo? @ruby rain level 1 !!!

[quaint zinc]

Guysss , can someone help me about running links on rvc cause ı keep getting the same error on googlecollab for different links

[rare goblet]

Ayo? @quaint zinc level 1 !!!

[quaint zinc]

How can ı fix this?

[thin stump]

You're supposed to clean and go through your whole dataset first before you use audio labels? Obviously it will sound dirty if you never looked at it

[bronze maple]

No, my dataset is super clean, it must do it for you too I suppose, I have the impression that rvc when it processes my dataset, it reduces the quality, check for yourself if it's the same

[bronze maple]

maybe it reduces the dataset to make training easier, I don't know, or Applio is just bad

[rare goblet]

Ayo? @bronze maple level 6 !!!

[thin stump]

I use RVC disconnected so I wouldn't know. You can try truncating instead with the aihub's setting

[bronze maple]

truncate? does this mean that we can directly put our dataset in the RVC Logs? but RVC must perhaps process my dataset in all cases necessarily, I don’t know

[thin stump]

https://docs.aihub.wtf/rvc/resources/datasets/#step-2-truncating-silence

[bronze maple]

ha but no that's the step before starting training in RVC, I'm really talking when you enter the access path and RVC processes your data and cuts it into small pieces of 2 to 3 seconds, then if you check in the archives of Applio/logs/YourModel/1-16k-wavs, there is your entire dataset reworked by RVC, and if you read, the quality is disgusting, and I think that the RVC works with that

[bronze maple]

Someone needs to check if it's the same for him.

[frank root]

rvc works on like 16bit 16KHz wav internally

or so I have been told

[sweet grove]

hey i need help with the google collab

when i paste the link to the voice model an error just pops up

cant send pictures but it underlines this part of the code "from Mega import Mega"

[sweet grove]

redid some steps and now some credentials are the issue??

[rare goblet]

Ayo? @sweet grove level 1 !!!

[sweet grove]

"---------------------------------------------------------------------------
JSONDecodeError Traceback (most recent call last)
<ipython-input-13-e47c5dcc25cb> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

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

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

[pastel oak]

Youre using an outdated colab

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

I'll link RVC-Disconnected for now but theres a few more like Ilaria on the left side, if you prefer to use any other feel free to

[half nymph]

i have the same problem, is there an alternative to this solution?? im new to this stuff

[pastel oak]

As I said, you were most likely using an outdated colab. Head over to the link to get started

If you are only interested in doing AI Covers and not training voice models, I would recommend using Ilaria RVC

[brittle wing]

Why is my model training so slow
Is it cause of the dataset being big...

[pastel oak]

Local or cloud?

[brittle wing]

Rvc DISCONNECTED

Public

[pastel oak]

Whats your batchsize, how long is your dataset, how long does one epoch take to train atm?

[brittle wing]

Maybe because other people R training at the same time?

[brittle wing]

Idk about batch size I'll send the dataset

One epochs takes about 3-4 minutes ...

It's annoying

[pastel oak]

You've defined the batch size on this step

[brittle wing]

https://drive.google.com/file/d/1-AGhKIZsOyls9mQpb1qyOt9FVNp1Mypb/view?usp=drivesdk

[pastel oak]

You can just tell me the minutes instead of sending me the dataset lol

[brittle wing]

Uh

Okay

[pastel oak]

There's already a few j-hope voice models out there, have you tried them before or you just want to do your own?

[brittle wing]

Where...

[clever nova]

i need help

which app

u guys use

for the voice models

im new heree

[brittle wing]

They're probably my old flawed ones

[pastel oak]

https://huggingface.co/4naluvs/HobiRap/resolve/main/HobiRap.zip
https://huggingface.co/elliyonline/jhopeRVCv2/resolve/main/jhopeRVCv2.zip

[clever nova]

please help me

[brittle wing]

jhopeRVCv2 is mine.

The trained on crepe one

[clever nova]

..

[pastel oak]

Can you define what exactly you want

Do you want to create voice models or just do AI covers

Knighty please chill lol

[clever nova]

ai covers

like the ones i dwonload

[pastel oak]

Youve not given me time to breathe to answer 💀

[clever nova]

i want to use them

[clever nova]

lol

[brittle wing]

It's my model

Hobirap is not

[clever nova]

which app

for pc

[pastel oak]

CHILL 😭

[brittle wing]

-colab

[azure marsh]

Suggestions for @clever nova

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[clever nova]

which is the lite st

like i mean without much load for my pc

[rare goblet]

Ayo? @clever nova level 1 !!!

[frank root]

need nvidia gpu to work

[clever nova]

oh

[frank root]

else can't use

[brittle wing]

Simply use ai cover generator

[clever nova]

without that we cant use?

[pastel oak]

You can use the online versions

[frank root]

yes, or you use online

[pastel oak]

theyre called Colab

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

[clever nova]

like i cant download the app and use?

if i dont have nvidia gpu?

[brittle wing]

I'm gonna calculate the dataset length and tell you....

[pastel oak]

correct

need nvidia gpu for app

[clever nova]

AAh fuck

so i cant use ti?

it*?

[pastel oak]

i already told you your alternative option

[frank root]

lolz

[clever nova]

whats that

[pastel oak]

💀

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

[clever nova]

ok thanks

il try it

btw

for the ai models are there language?

or we can use any language

[brittle wing]

THE MODELS CAN AING IN ANY LANGUAGE YOU WANT

Sing

What's being copied is the voice not the language

[clever nova]

sing*

like i cant use it in my mic or somethin?

[pastel oak]

It seems like you have a wrong understanding of how this works

You download voice model (example: Ariana Grande)
You have acapella of a song you want to convert to Ariana Grande voice
You open Ilaria RVC and follow the steps in the guide
You have new acapella song of Ariana Grande

[clever nova]

ohhh]

[pastel oak]

You can also sing the song yourself then convert it into Ariana Grande or anyone else you got the voice model of

[clever nova]

so i cant

use this?

na i mean like ingame i cant imitate another voice?

[pastel oak]

That is a whole nother thing

Theres an app for realtime voice changer, you should have just told me what you wanted from the start

[brittle wing]

Droid

H

[clever nova]

.

thats what i want

[frank root]

kek

[pastel oak]

Ok first question what GPU do you have, it can also work with AMD if you have that but I need to know what kind

[clever nova]

i have intel

[frank root]

yes but no

[pastel oak]

Intel UHD Graphics?

[frank root]

the sounds will differ, will mostly sound bad

like it doesn't say vowels the same way

[brittle wing]

This new dataset contains samples from his new solo album & it's cleaner.

That's why I'm retraining

[pastel oak]

Ok nvm just do online version

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

The guide on setting it up:
https://docs.google.com/document/d/e/2PACX-1vTIceEcBfS6Zqolv_QEysrFfI_EJikPxozWptP_EjkpLVl-l-gdo-ijBonQMTviAHEYm5emmd9k9TdC/pub

Anything else, such as uploading voice models and settings, check out
https://rentry.co/VoiceChangerGuide

Already a heads up, you can skip the "[Optional] Upload a voice model (run this before the voice changer)" since it doesnt work, you can upload it later when the GUI opens

@clever nova most importantly PLS DO READ THE RENTRY GUIDE, you will especially need Step 6: Virtual Cable to make it work on games

If any questions come up, ask in #🔍│help-w-okada

[clever nova]

ok

[rare goblet]

Ayo? @clever nova level 2 !!!

[grim nova]

ilaria rvc doesn't works

it says error

[brittle wing]

I'm calculating the dataset length through a calculator

[clever nova]

u guys hiring mods for the server?

[frank root]

lol

lmao even

[clever nova]

wht?

[frank root]

you just joined lol

[pastel oak]

It would probably be best to first have a look around and get to know everything before applying

[frank root]

based

[grim nova]

question

the acapella file has to be WAV?

in iralia rvc

[frank root]

anything common should work

[pastel oak]

I dont think so

[frank root]

like mp3 wav m4a or all that stuff

[clever nova]

..

[frank root]

..

[clever nova]

where to apply?

im pretty bored

im free from like 11 am to 3am

not pm

thats like 16 hours

a day

[frank root]

https://tenor.com/view/cat-wtf-oh-hell-nah-gif-4183712492046775434

[pastel oak]

Learn about RVC, w-okada, stable diffusion etc. if you want to

[clever nova]

..

[pastel oak]

Other question

What can you do as a mod that you cant do right now?

[clever nova]

seriously

[frank root]

what

[rare goblet]

Ayo? @frank root level 5 !!!

[frank root]

silly cat gif

[clever nova]

help maintaining rules

make sure everything is right

[pastel oak]

I am not sure if the people overseeing applications will trust you right away considering you have only joined 5 days ago. You can try, but I personally think there's no point but I wont stop you if you dont care

[clever nova]

ok

where to apply

[pastel oak]

do /jointeam , select which role you want to apply for, hit enter again and it will open up a channel ¯_(ツ)_/¯

[clever nova]

ok

[brittle wing]

About 1 hour & 18 minutes!!!

Uh

[pastel oak]

So I dont know if the colabs training speed depends on your internet or not

likely does, idk

[brittle wing]

It's still slow...

[pastel oak]

in that case, 3 minutes might be realistic here but I have never trained on colab so i could be off

It should probably be a bit faster

[brittle wing]

Uh

How do I train faster

[pastel oak]

buy a gpu and train locally

4head

[brittle wing]

How many epochs for a 1 hour and 18-20 minutes dataset?

[pastel oak]

me on my 3060ti batch size 8 can train an epoch in 25-40 seconds on a 25-30minute dataset

[shrewd coral]

bro

i need help

[brittle wing]

Hypothetically how many

[pastel oak]

just ask

[shrewd coral]

what's the site to make ai covers

[brittle wing]

How many epochs should the model be hypothetically

[pastel oak]

probably 100 in an hour

oh

w8 ill answer

[pastel oak]

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

[brittle wing]

Nope I know it completes 100 in an hour

[pastel oak]

Every voice model is different so it depends when it starts overtraining

[brittle wing]

I'm asking according to dataset length

[pastel oak]

you use tensorboard to estimate that

can be 250, can be 600 for all i know

[brittle wing]

I know but hypothetically how many according to dataset length

[pastel oak]

dont know

[brittle wing]

800?

[shrewd coral]

💖

[pastel oak]

I really dont know

[brittle wing]

Why is it so slow then with T4 is it because of the dataset

[rare goblet]

Ayo? @brittle wing level 8 !!!

[pastel oak]

Probably

[rare goblet]

Ayo? @pastel oak level 29 !!!

[pastel oak]

or Internet speed?

[brittle wing]

Uh

Could be

[solid mica]

hi

where can i download okada voice changer? 🥰

[pastel oak]

https://rentry.co/VoiceChangerGuide

Everything you need is on here depending on your GPU and all

[onyx rampart]

uhh

So I have just finished training a model, and it has this disgusting tinny metalic noise

oh god

it's horrible

I should mention that my vram was absolutely dominated completely by having a batch size of 15

I have 12gigs of vram so I have no idea if that's my issue

but tbf it did the whole 200 epochs suspiciously shortly

[plucky socket]

i duplicated this space and have been using it mainly for inference, but i'm paying for gpu use for the space now and i realized last night that i can actually train voices on it now as well. i was able to do one last night and inference it but i couldn't figure out how to actually export the voice model/index externally for later use. and the space doesn't seem to have saved the model anywhere. does anyone have any suggestions or is maybe doing what i'm doing with this specific space? https://huggingface.co/spaces/r3gm/RVC_HFv2

[onyx rampart]

I believe I was at 130 at 20 minutes

[onyx rampart]

I'm not entirely sure you can actually download models from this cloud fork

I know you can definitely download audio outputs though

[plucky socket]

yeah i was gonna say i can download audio outputs with no issue so i feel like there has to be some way to download the pth and index files

[rare goblet]

Ayo? @plucky socket level 1 !!!

[plucky socket]

idk i haven't messed with ai rvc stuff in months so i'm kinda relearning everything, i miss the old easy google colabs rip

[onyx rampart]

Yeah but they were gonna be wiped out sooner or later by google

they do not like making it easy to use such powerful tools

But then again it kinda is really easy if you have enough money

[stuck yew]

https://huggingface.co/sail-rvc/adamlevine how do i download this? it says 1 download alst month but all i see are seprate downloads for each file

[vast fractal]

how do I setup this rvc program, I got tired of the old one so I decided to get this one but have no idea on how to set it up https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

[onyx rampart]

I can help

hop into a vc so I can walk you thru it

I'm in https://discord.com/channels/1159260121998827560/1159291174926942268

[glass oyster]

Yo what is everyone using for an easy song conversion today?
kits.ai has a file size limit and I can't upload my model properly

[onyx rampart]

the rvc ui

it's easy to use bor

bro*

[glass oyster]

I like kits for the quick youtube conversion

[onyx rampart]

Yeah but I have this tampermonkey addon that can let me download the audio's off of youtube videos

[rare goblet]

Ayo? @onyx rampart level 5 !!!

[glass oyster]

With the vocals already separated?

[onyx rampart]

No you would have to run that either through https://ultimatevocalremover.com, or through the rvc webui panel that uses the exact same model on uvr5 to seperate tracks and vocals

but there are websites that work

(and are free)

[glass oyster]

yeah, i've had this setup before but it was too much of a bitch for every single song and track

[onyx rampart]

But I find you get the best quality using uvr5 because you can mix and match models to further improve vocal extraction

[glass oyster]

really dont want to jump through 10+ steps for a single meme cover

kits is really convenient for that

just paste a link and leave it

[onyx rampart]

Yeah that doesn't exist without a price

@vast fractal You there man?

[stuck yew]

how do i fix

---

JSONDecodeError Traceback (most recent call last)
<ipython-input-15-39369b45582a> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:

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

JSONDecodeError: Expecting value: line 1 column 1 (char 0) when using google collab https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=OVQoLQJXS7WX

[onyx rampart]

There may be open source projects you can find online. But it's gonna be super finicky

[bronze maple]

Is there anyone experimenting with RVC? I just found something that destroyed the quality of the dataset in Appliov3 during training

I would like to know if it does the same to you

[onyx rampart]

You might wanna reset your session, unless you have done and have already come across this error

[stuck yew]

yes i have done that twice

[onyx rampart]

multiple times*

[frank root]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[onyx rampart]

Okay, I would use one of these fine selections that hexos has kindly provided

[stuck yew]

ok

[onyx rampart]

These are known to work well by the community

[frank root]

looked it up, you are using a old borked one

apparently

[stuck yew]

oh

yeah i got it from a yt tutorial so makes sense

[onyx rampart]

How old is the video?

[stuck yew]

agu 20, 2023

aug

[onyx rampart]

Oh I can't send my funny emojis

Well that ruined that

Hey hexos, you have any tips for my problem that I posted up earlier?

Kinda got flooded out by chat

[frank root]

idk, when I trained models it worked

just that batch size 6 was almost maxing my 8GBs of VRAM

[onyx rampart]

Yeah I think it may because I went a little too crazy with the batch size

And I believe that might be the issue

Dude I couldn't even train with 8GBs of vram

[frank root]

huh

[onyx rampart]

My 2060 was real crappy

wasn't even from any decent brand

[frank root]

how much did you push them settings

8 should be like fine

[onyx rampart]

I didn't, even on really low batch sizes like 3

it just came out sounding all tinny and metallic again

[frank root]

https://tenor.com/view/what-are-you-talking-about-weird-look-innocent-gif-19323620

embed failure

comedic moment

[onyx rampart]

which is why I suspect that topping out my vram was the issue

comedic failure of embedding type beat

[frank root]

i would be surprised if it did corrupt anything

at worst it crashes if it don't have enough

[onyx rampart]

Oh no it never crashed

[frank root]

or becomes slow because nvidia made it use system RAM as fallback

[onyx rampart]

just was VERY topped out

like running at it's top clock

[frank root]

oh that's normal

[onyx rampart]

I mean I read that RVMPE has this issue with metallic sounding voices

Sometimes ( not always! ) rmvpe-trained models have a metalic overlay-tone to it and using rmvpe on such models, further enhances the effect.
Yeah, this is what I read

oop RMVPE**

welp I'm retraining using mangio-crepe and cranking it down to 10 on the batch size

[brittle wing]

What trains faster?

Rvmpe or mangio crepe

My model is training way too slowly

The gpu is overloaded probably

[vast fractal]

it is

[brittle wing]

Why

[vast fractal]

training requires a lot of resources

[brittle wing]

It's not dataset being one hour and 18-20 minutes then

[vast fractal]

it might also take a little long for a longer dataset

[brittle wing]

Nah isn't it cause other people are training at the same time

[brittle wing]

1 hour & 18-20 minutes

[vast fractal]

probably your large dataset then

are you running it local or public

[brittle wing]

Rvc disconnected

Nah I think cause other people are training too

[vast fractal]

probably if it is public then yea

[brittle wing]

It's annoying

All they do is occupy

[vast fractal]

yea

[brittle wing]

Are they training now...

[vast fractal]

probably

[rare goblet]

Ayo? @vast fractal level 3 !!!

[brittle wing]

The model won't be finished until tomorrow my goal was to finish it for 2 days!

[vast fractal]

yea it will take a while

[brittle wing]

How do I train on a private GPU on mobile

[odd shale]

You can't.

[brittle wing]

But it trains slowly and all that happens is that I run out of GPU

[odd shale]

Are you trying to use Colab on phone?

[brittle wing]

And I can't even finish the first 200 epochs of it

[brittle wing]

Tablet

[odd shale]

But you're using Colab on Tablet?

[brittle wing]

Does it matter on what device

[pastel oak]

Yes theyre using colab, or trying to at least

[brittle wing]

I'm struggling

It's very slow

Is astralabs an alternative

[pastel oak]

The question Leo is asking is a yes or no question before they can try to continue helping

xd

[brittle wing]

Yes

[odd shale]

Welp, normally you should be able to select T4 on Colab.

Altho i find pretty unconfortable to try training a model on phones or tablets.

[brittle wing]

I reconnected to it now

[brittle wing]

Don't ask me 😐 it's a torture

[odd shale]

You got no computer?

[brittle wing]

Now it's faster

[brittle wing]

I don't own it

[odd shale]

In case of anything, you should try reading our guides for troubleshooting.

[brittle wing]

It's training faster after reconnecting

[odd shale]

https://docs.aihub.wtf/

[brittle wing]

How do I find them what command

[odd shale]

There you have.

[brittle wing]

I know these

-guide

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[brittle wing]

It's the dataset length I think

Now it's faster

[odd shale]

Everything you need it's on the guide i gave you.

[brittle wing]

I know all of It

I accidentally added the same sample two times in the dataset

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[hot lagoon]

can someone send a link to diarisation ai?

[brittle wing]

Why does the MDX colab keep disconnecting everytime I try to isolate vocals?

[modern zodiac]

Whats the preferred rvc atm applio or Ilaria?

[wispy burrow]

So far I prefer Ilaria RVC Beta (local) over Applio RVC (Local).

[modern zodiac]

Say less, i've been using mangio and then did some reading here and WHAAT people just keeping making ****

[modern zodiac]

Any particular features that stand out to you personally compared to what Applio has to offer?

[rare goblet]

Ayo? @modern zodiac level 1 !!!

[wispy burrow]

Applio is the one with the most features.
but it's also the most unstable one.
Its deffinitely the version I get the most errors on.

I like Ilaria RVC more because it is simpler and straightforward.

[modern zodiac]

okok, i mean stability is good, i'd be pissed if something crashed while i was training

so if we're simply talking best at training in terms of best method is it pretty much similair between the two

[hot lagoon]

?

[spring lagoon]

is there samples you can here in find models?

I mean voice models

[shadow rover]

Covergen colab needs a nvidia card even though it was working fine without one how do i fix it?

[proud elbow]

make sure the runtime is using T4 gpu

[quaint pollen]

guys

can someone hel[

how do i fix frequent erros occur. Please ceck if the model of the framework being targeted is loaded

[shadow rover]

It still doesn't work plus it says "Cannot connect to gpu backend."

[rare goblet]

Ayo? @shadow rover level 1 !!!

[proud elbow]

wait tomorrow or switch to another account

[quaint pollen]

how to run illira rvc

#✨│ai-help

[onyx rampart]

defo rvmpe

But it has it's cons too

It is fast but has less vocal detail, unlike mangio-crepe which is much much slower, but way higher in quality

[fallow pivot]

Hi guys i have a problem with easyGUI- 3/09/24, at the final step (5 Download Model) with an instant error "FileNotFoundError: [Errno 2] No such file or directory: 'assets/weights/name' ".

[onyx rampart]

I have been training my own voice for the shiggles with my own singing, and this is my results with magio-crepe at 200 epochs

My voice is weird however, I have to always use -12 pitch becaus my voice is very uh

deepish?

You can tell the pitch is too low at points, but it doesn't sound like me at all without lowering the pitch by an octave, which is how I sing naturally due to being incapable of reaching higher notes

also the autotune sound is coming from the song, it's crazy how you don't notice how much musicians change their voice in post, but it is very obvious over my own voice

[modern zodiac]

whats the best model for training in yalls opinion for highest quality.. assuming that the dataset is mildly trash

[onyx rampart]

In my exerience, (which is little), magio-crepe is very good using the mangio crepe fork

ignore the extra spaces, my bluetooth keyboard sucks ass lol

[modern zodiac]

thanks, personally been using rvmpe however some results have been meh

[wispy burrow]

If you mean the Extraction method, RVMPE is usually considered the best overall.

[hot lagoon]

can someone send a link to diarisation ai?

[modern zodiac]

ye, brains all mushy today from all this reading

[wispy burrow]

Well, you need a clean Dataset if you want good results.,

[proud elbow]

rmvpe is good and stable even for average quality, but crepe (w hop length) can be better for very clean audio yet might be more sensitive to little noise

[modern zodiac]

yeaa, that's the tricky part for the project I'm working on atm

@wispy burrow you seem very knowledgeable, ive read some great info here: https://docs.google.com/document/d/17fjNvJzj8ZGSer7c7OFe_CNfUKbAxEh_OBv94ZdRG5c/edit .. any tips you could personally bestow for removing trash noise

[rare goblet]

Ayo? @modern zodiac level 2 !!!

[onyx rampart]

yeahh, a somehwat decent microphone is required when you're making your own datasets

But I have a razer seiren mini and that is plenty good enough

Oooh I have a new role c:

[wispy burrow]

Oh, never sseen that guide before.

[modern zodiac]

if only twas that easy 😢 im working on making a model of sassy the sasquatch from the biglezshow if you've ever heard of it

[modern zodiac]

some lovely stuff, whole lotta knowledge in there

[onyx rampart]

defitely a lot to keep track on that's for sure

definitely*

[onyx rampart]

of coouuurse mate

[wispy burrow]

Gonna take a while to read all that.

[modern zodiac]

maan see, thats the easy part, i've already got all the vocals from all episodes and its split beautifully, most noise removed but there's still imperfections which render the dataset poop

[karmic vessel]

anybody elses thing just shut down?

[rare goblet]

Ayo? @karmic vessel level 1 !!!

[onyx rampart]

If you want, I don't mind trying to clean the dataset and train it all up for you if you'd like?

free of charge of course

I don't mind helping the cause, especially sassy man

[onyx rampart]

Context?

like, any error codes?

That is uncomfortaly brief xd

[karmic vessel]

yeah umm it said it had issues initializing?

[onyx rampart]

send over the log

let's have a cheeky peak

[fading drum]

colab?

[onyx rampart]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[onyx rampart]

If you're using colab, I highly reccomend one of these to use

recommend*

[modern zodiac]

appreciate the offer, I think for me it's more satisfying to go through the whole process, do you have any suggestions or methods you personally use for cleaning your datasets?

[onyx rampart]

well if you have uvr5, using the MDX23C can filter out pretty much anything

https://ultimatevocalremover.com

[rare goblet]

Ayo? @onyx rampart level 6 !!!

[karmic vessel]

OH I see what happened

it reset the audio as well messed up my inputs and outputs

[onyx rampart]

And this is what it should look like if you choose the right options

[onyx rampart]

rip dude

Make sure to use either .wav or .flac because lossless audio files are much better for training

I personally use .flac because that's just what I've been using

kek

[wispy burrow]

Then use DeEcho-DeReverb.
And finally, DeNoise.
those 3 should get one pretty far.

[modern zodiac]

oh rough 😢 ive already done all that haha.. thought maybe you had some secret wisdom

[modern zodiac]

daang, I need some nasa level shit haha

[onyx rampart]

yeah but you can't use MDX23C with those in ensamble though?

They dissapear from my list when I choos ensamble mode

Again, ignore my ass keyboard lmao

disappear*

That one was a genuine spelling mistake tho lmao

[onyx rampart]

Anyone should be able to do that locally though

[wispy burrow]

I use them one at a time...

[onyx rampart]

oh you poor soul

[modern zodiac]

im with ya there, mdxvocft, dereverb and then denoise

[onyx rampart]

yeowch

[bronze maple]

do you know where the Tensorboard is in Mangio RVC?

[modern zodiac]

https://www.youtube.com/watch?v=P0M7PAsG1fk this should help

[wispy burrow]

It's the tbL bat file

[bronze maple]

where is ? 😂

tbl.bat files ?

[wispy burrow]

Ugh...forgot those weren't included in mangio's zip...
We needed to download those separately...

[bronze maple]

But no ? how do they do it then wtf

[wispy burrow]

Now where is that old guide that had the download link for those...

[proud elbow]

use tensorVENV.bat from the guide https://docs.aihub.wtf/
and it works for any RVC forks

[wispy burrow]

Even better.

[bronze maple]

You are the best

[echo lava]

how long should RVC be processing a 6 minute mp3 with CPU?
R5 4500U
8gb ram
no EGPU
answer: npy: 91.73318481445312s, f0: 1145.1132094860077s, infer: 425.0613353252411s

im using crepe

[faint moon]

i face real time delay rvc voice changer plz help how to fix

[echo lava]

nevermind it just finished lol

[spark mirage]

How much did it take?

[echo lava]

27.6 minutes

[rare goblet]

Ayo? @echo lava level 1 !!!

[spark mirage]

Wow

[rare goblet]

Ayo? @spark mirage level 1 !!!

[echo lava]

yep

[spark mirage]

Quite long

[echo lava]

when i had my 1050ti it was like 30 seconds

yeah

[spark mirage]

I want to be able to convert female sound to male, but I guess it's impossible to run it cheap

[hot lagoon]

diarisation ai where can I find it if anyone has a link for it

[onyx rampart]

adjut setings, or, stop using a gtx 750

[onyx rampart]

I mean there's guides

[spark mirage]

Do you have links for them

To run it cheap?

I have tried to degrade the pitch to get the effect, but results were poor, so I thought I have to use ai

[spark mirage]

?

[wet wren]

the process data button isnt working, it just flashes once and then doesnt follow up with any output info

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[wet wren]

i also dont know if it was just taking a while, im training it off of one audio file and its 40 minutes long

[acoustic garnet]

it should take a while

I remember 40m datasets preprocessing at least 10-15m and i would just come back to checkup on it later

[misty elk]

How do i create a dataset

[acoustic garnet]

its just audio files. record yourself or rip audio from whatever source you want.

[wet wren]

will i know if its processing or is the output info just gonna be blank until its actually finished

[rare goblet]

Ayo? @wet wren level 2 !!!

[acoustic garnet]

just check task manager if your system is doing something

[rare goblet]

Ayo? @acoustic garnet level 7 !!!

[wet wren]

alright, and one more question in the mangio folder do i just make my reference audio its own folder or put it in datasets or does it not matter

[acoustic garnet]

you usually put a folder path in the text box containg all the audio files you want to use. likely there is already a dataset folder

[wet wren]

i'll just use that then, thank you

[misty elk]

So uhh how do i use easy gui

[rare goblet]

Ayo? @misty elk level 1 !!!

[acoustic garnet]

#1159513888199540817

[acoustic garnet]

can you screenshot the webui?

[acoustic garnet]

you cant have spaces in the folder path

maybe you can put "" around the folder path

it cutt off on new folder

[wet wren]

oh found the space, now its just new errors

[acoustic garnet]

is all your audio files in the datasets folder?

it cannot be in a subfolder

[wet wren]

yes its just one file

[acoustic garnet]

hm

[wet wren]

should i just try reinstalling?

[acoustic garnet]

mangio is not updated anymore i think?

i've ditched it for just mainline or applio

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

[modern zodiac]

I was thinking of using applio or Ilaria, whats are your thoughts on applio?

[acoustic garnet]

it was fine when i used it. it was half a year ago

i've never used ilaria

🙂

[wet wren]

how do i even install this i always forget everytime i open a github link

[acoustic garnet]

on the right side it says releases

[modern zodiac]

thats what i'm stuck between, cause mangio is deprecated

[acoustic garnet]

use the hugging face download link

there should be a installer script

[acoustic garnet]

applio i think has the easiest installer

[wet wren]

where would i find that

[acoustic garnet]

-rvc

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[acoustic garnet]

-fork

hm

-local

[azure marsh]

🖥️ Local stuff

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

Need some help? 🤔

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

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

[wet wren]

there it is

fancy installer woa

[sacred crow]

i download phineas ferb but it said file not found when i try to save it

in voice.ai

[plucky plover]

Hi,
I've got an RVC v2 RMVPE voice model
I want to make a cover for a song with this model

How to do that?

I have .index and .pth files

[pliant surge]

Does anyone have a voice model of ayaka jp, please let me know

[plush compass]

What do you all use as a local RVC? I changed my computer this time, but it's my first time using local, so I don't know what to use. Applio??

[pastel oak]

Never tried applio personally, I use Mainline

[plush compass]

thx

[plush compass]

By Mainline, do you mean Mangio-RVC?

[pastel oak]

Nope, this one https://docs.aihub.wtf/rvc/local/mainline/

It is a bit faster than Mangio in performance from what I know

[plush compass]

Thank you. Have a good day

[pastel oak]

Everything you need is on the guide

https://docs.aihub.wtf

[onyx rampart]

youtubeeeeeee

Literally the first thing that shows up is very understandable guides

and there's many guides in this server too

https://www.youtube.com/watch?v=gtjCmJ_6d0k

that guide is for colab, which means you don't need a gigachad gpu to live life

[onyx rampart]

Uh oh

I just tried training with rmvpe

again

And it has proceeded to shit on me again

wahoo

This is what it sounds like

you gotta skip forwards a bit

damn 1:30 hits hard

absolute bop

[old oracle]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[brittle wing]

anyone knows how can i get this voice https://www.youtube.com/watch?v=0dgdVRndfqg at 32:24

[onyx rampart]

bruh

you just lower the pitch in post

If you wanna do that I can reccomend you the perfect too]

[brittle wing]

so how can i make it w okada ?

[onyx rampart]

There's no point doing something so simple with AI

you can do it with normal voice changers

like this one

https://clownfish-translator.com/voicechanger/download.html

All you do is just turn the pitch down ._.

[slim geyser]

Yeah shows have been doing that for decades lul

[whole elm]

hello guys can i ask u how i can make my voice work properly on that app?
https://imgur.com/a/3nIwZfF
the voice is 2 much similar to mine

[proper shale]

mess with tune

[brittle wing]

?!?

1 hour 18-20 minutes dataset overtrained at 100 epochs?

[pastel oak]

thats not enough to determine it imo

[brittle wing]

I think so too

Admin told me to keep training

[pastel oak]

like i go 15k + steps into overtraining before making sure whatever i got was the OT point

and youve just crossed 15k as a whole

[brittle wing]

It's not overtraining

?

[pastel oak]

I am saying is that to determine OT you need to keep training past that point to make sure it really is the Ot

you cant make it past a point and immediately say yes thats the OT point, you need to keep training for easily 30+ mins before you can come close to making a decision

[brittle wing]

...

[silk nimbus]

guy help when i press start and say something, theres alway a second voice. How how i turn it off?

[fleet charm]

someone now when i click start_http it like open some windows and close it in the same time like on secone is here and after 2 sec close

[nova gull]

does someone know why i cant hear myself

or anything

I can't hear my voice can someone help me with that

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[muted leaf]

anyone know why my gpu jumps to 90% usage after tabbing into cs2. my gpu is at 10% while being in esc menu or not in the game. is there a setting to change that or nah? otherwise my voice changer is lagging

[astral perch]

any way to remove footsteps from audio?

[mortal lodge]

can anyone help me with tensorboard not showing any data even though the training appears to be going?

[winged fulcrum]

so if a model is set to private, will it be useable within the rvc software itself still if someone has the link?

[mortal lodge]

nvm got it figured out

[silk hamlet]

is not an accepted origin. (further occurrences of this error will be logged with level INFO)
Did anyone had this error with w-okada's Voice Changer | Colab on google colab?

[brittle wing]

so for gpt-sovits people,maybe soon we would get a document on how to use the gpt-sovits colab?

[thin stump]

You only share the huggingface link with a few select people while it's public then you could private it

[thin stump]

you only need the index and pth tbh. I don't know why others would insist on the huggingface link itself

[sinful nexus]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[visual flint]

The save to drive button on INSTALL RVC step - what does this do? Atm i believe i have to do both steps every time i want to start a new session, but if i do save to drive will i be able to skip that step on future uses?

[proven hill]

no

[visual flint]

Whats the point of it then?

[half cove]

to save trained models to drive automatically

[visual flint]

Ah ok, thnx

[silent iris]

How i put a model in the voiceai?

[visual flint]

What directory/folder does the load_models_from_drive option take from?

[silent iris]

wdym?

[half cove]

logs and weights iirc

[half cove]

if your talking about voice.ai we dont support that here

if your talking about rvc

https://docs.aihub.wtf/

here are our guides

[silent iris]

Oh ok thx

[rare goblet]

Ayo? @silent iris level 1 !!!

[silent iris]

Srry

[visual flint]

So if i save to drive, run the UI, download a model from the UI and use it, the next time i start a session it should already have the model downloaded? I take it the model is downloaded as some obscure files rather than a .zip

[half cove]

ngl i dont know if that part of the code works (i didnt write the load to drive only save to drive)

but it should load folders for retraining and also weights

so models used should be saved

i think

[bronze leaf]

when training a model, which option is the best for best quality?
pm, harvest, dio, rmvpe, rmvpe_gpu ? i have a nvidia rtx gpu

[visual flint]

ok thnx for the help. i dont suppose there's any way to speed up/shortcut the loading process? (mostly the "install rvc")?

[half cove]

rmvpe/rmvpe_gpu both use rmvpe but gpu uses gpu ofc so its faster

[bronze leaf]

is this for quality better then harves?

[rare goblet]

Ayo? @bronze leaf level 1 !!!

[half cove]

i could add saving prerequirements to drive but that yeets your storage

[half cove]

yes

[bronze leaf]

ok thanks

[visual flint]

Fair enough nvm then. Thanks again!

[rare goblet]

Ayo? @visual flint level 3 !!!

[sinful nexus]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[brittle wing]

https://huggingface.co/spaces/IAHispano/Applio
Hello, how do i make my own Cloned Voice with this?

[visual flint]

I've just downloaded a voice model zip from the server, what is the .index file for?

[tame mica]

accents

its not really accurate though

[visual flint]

so i could import just the .pth file into colab rvc and it would work fine?

[tame mica]

ye?

why colab

you don't need good specs for local inference

[visual flint]

cos i want to be able to use it on mobile

i like the GUI as well, only thing is it takes a while to load but still a great free option

[tame mica]

fair enough

[sinful nexus]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[brave iron]

does anyone know any softwares that i could use to use these ai voices

to like voice change my voice

[visual flint]

Google Colab RVC. Can be found in this server somewhere

[low shard]

for ai covers/audio to audio voice conversion check https://docs.aihub.wtf

else for realtime wokada voice changer

-rt

[azure marsh]

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

[low shard]

check the first guide

[nova gull]

i cant hear myself

anyone help??

[fiery wraith]

problem

it infinitely processes

on the huggingface rvc v2

[quaint ether]

help

im using the tortoise rvc inference thing

and i got this error

Possible latent mismatch: click the "(Re)Compute Voice Latents" button and then try again. Error: Workspace can't be allocated, no enough memory.

(btw not even 25% of my ram is being used i got 64 gb)

[visual flint]

So ive got my own voice model on google drive and want to import it to Colab. I paste the link as shown and hit download, it says it's done, i go into the main page and hit refresh to use the model but it never shows up unless i download one via the second search menu. Any ideas?

@half cove ?

[half cove]

which colab

[vast fractal]

i was trying to further train my model to 200 epochs then my rvc just said this: raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs768NSFsid:
size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 256]) from checkpoint, the shape in current model is torch.Size([192, 768]).

[split gulch]

I got the same error as you get when I was trying to train my model.

[vast fractal]

yea

[proper nebula]

hey is any one online

i need some help

im trying to input a voice model i found on here into the google colab on the second section but it keeps saying error

[rare goblet]

Ayo? @proper nebula level 1 !!!

[proper nebula]

anyone know how to fix it?

[odd shale]

Send screenshot.

Also, send a link of the model you wanna use to check.

[proper nebula]

thats the link

[odd shale]

Austrian Painter?

[proper nebula]

ehehehehe

[odd shale]

Wait.

Which colab are you using?

[proper nebula]

um

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=OVQoLQJXS7WX

that one

[odd shale]

I think that colab is outdated tbh.

[proper nebula]

oh rlly

is there a new one

?

[odd shale]

Well, i suggest you to read the docs.

https://docs.aihub.wtf/

[vast fractal]

i was trying to further train my model to 200 epochs then my rvc just said this: raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs768NSFsid:
size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 256]) from checkpoint, the shape in current model is torch.Size([192, 768]).

[wispy burrow]

I am this getting this error when start training with Ilaria RVC Beta 2.

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

[brittle wing]

File "<ipython-input-7-d469f36c734e>", line 5
import oshttps://huggingface.co/Bruh110/YNWMELLY/resolve/main/YNWMELLY.zip?download=true
^
SyntaxError: invalid syntax

im gettin this issue

[low shard]

Which colab are you using

[brittle wing]

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=OVQoLQJXS7WX

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

idk how to access the other ones

[low shard]

Thats outdated

[brittle wing]

whats the newest colab

[low shard]

You can check them in the cloud folder inside the rvc folder of our docs https://docs.aihub.wtf

Theres also guides of them there

[brittle wing]

can u just link me a working colab i know how to do that rest

[low shard]

I mean if you really say so i can suggest one from the docs https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing

[brittle wing]

Traceback (most recent call last):
File "/content/Ilaria-RVC/app.py", line 1, in <module>
import subprocess, torch, os, traceback, sys, warnings, shutil, numpy as np
ModuleNotFoundError: No module named 'torch'

[low shard]

Click runtime, disconnect and delete runtime, and re run everything again, its js an error of colab being colab lol

[brittle wing]

now im getting an error when converting

[low shard]

Show exactly what error

Like the output in colab

[brittle wing]

yes

showing an error when i convert from my audio to ai

[wispy burrow]

He's asking for a screenshot of the error that is shown in output.

[brittle wing]

yh i gave up

ima try again later

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[vast fractal]

i was trying to further train my model to 200 epochs then my rvc just said this: raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs768NSFsid:
size mismatch for enc_p.emb_phone.weight: copying a param with shape torch.Size([192, 256]) from checkpoint, the shape in current model is torch.Size([192, 768]).

[lime otter]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

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

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

[ivory orbit]

I NEED HELLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLP

[rare goblet]

Ayo? @ivory orbit level 1 !!!

[ivory orbit]

Shut up

Can someone help me

I don't know why but um

I fucked something up

[honest rivet]

Hi, i can't see or hear a volume output at the Voice changer client. Input and Output are correct. Does someone know what the problem could be?

[spring turtle]

how do i get bs reformer in UVR 5?

[fair nymph]

Anyone know how I use a voice for discord?

[stiff grotto]

Go web bat is not launching

Not launching the interface on the browser

[sinful chasm]

hey guys why is the client not working for mac

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[plush compass]

What algorithm do you all use to remove echo and reverb from UVR?

[carmine hearth]

https://youtu.be/IS_SPQVv5iY?si=eM-fcoJe5Kg06rAf
https://rentry.co/VoiceChangerGuide#virtual-audio-cables-mandatory

[wispy burrow]

VR Architecture
UVR-DeEcho-DeReverb

[proud elbow]

either of:

• MDX dereverb (only works for stereo reverb) then UVR deecho
• UVR deecho dereverb (increase aggresion till no reverb remains)

[long jungle]

guys why isnt my voice changer workig

i think its because the index isnt loadiung

anyone help me please

[tame yew]

You'll have to be more specific on how it isn't working

[long jungle]

like im talking into it

but like it just outputs my normal voice even tho i have it on

the voice changers on but not working basically

[tame yew]

Sounds like the output isn't routed correctly

[rare goblet]

Ayo? @tame yew level 1 !!!

[proud elbow]

click the passthrough button if it's red

[long jungle]

i did have pass through on

[rare goblet]

Ayo? @long jungle level 1 !!!

[long jungle]

what should it be outputted to cus im almost sure i have it the correct way

i watched numerous vids

[tame yew]

That should fix it

[long jungle]

it doesnt though

[proud elbow]

that passes your original voice, set it off

[long jungle]

but how will i be able yo know if the voice changer works

[pastel oak]

use "monitor" to your headphones

[long jungle]

it works now thanks but

why does it sound kind of robotic

[pastel oak]

send a screenshot of your voice changer and tell me the name of your GPU

[proud elbow]

set index 0, or it might be model quality or limitation of the onnx directml

[long jungle]

i literally love you people who help so much like why are you guys so helpful i swear to god god bless you eid mubarak

[pastel oak]

bereketler eid mubarak

[honest rivet]

.

[pastel oak]

RVC or w-okada

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[marble seal]

hey guys, im new to this rvc thing. im not really good at tech so does anyone have a guide for the ai voice changer?

[honest rivet]

Where can i see that? but i think RVC

[rare goblet]

Ayo? @honest rivet level 1 !!!

[pastel oak]

send a screenshot of your voice changer

[pastel oak]

thats wokada btw

Anyway I assume you have an AMD gpu. if you have nvidia you downloaded the wrong version

you havent selected an actual gpu yet. Open task manager, go to performance and check which number your GPU is (most are gpu0 or gpu1) then select it on the right side

Output has to be the virtual cable (depending on if you downloaded vb-cable or vac, it will be called "CABLE Input" or "Line 1 (Virtual Audio Changer)"

for monitor to hear yourself you can put in your primary headphones or singe i notice you use an audio interface its probably the main out 1/2

and then on discord or video games you select the virtual cable as your INPUT so its reversed

[honest rivet]

still doesn't work.

[pastel oak]

move the s.threshold tiny bit to the right

[honest rivet]

doesn't work either. maybe the model isn't loading?

[pastel oak]

can you enable passthru on the top and talk just to see if your mic is working normally? it will play back your normal voice. if yes, then turn it off again

[honest rivet]

can't hear anything

[pastel oak]

the problem is your mic then

[honest rivet]

it works everywhere else

[pastel oak]

¯_(ツ)_/¯ wokada is not detecting input

unplug replug your interface or check if you really have the correct input, also make sure your mic is not currently being used by another software

since the AMD version of wokada is pretty bad it could also just be that it thinks the mic is being used by the interface (even though that makes no sense) but it might treat it as another device accessing it

[honest rivet]

where can i see this? Thank you for answering so many questions, i'm quite lost^^

[pastel oak]

Idk off the top of my head

But tbh wokada would crash if its being accessed by another device so i dont think its that

Maybe try a different input of where your mic is plugged in, or use a different mic

[rare goblet]

Ayo? @honest rivet level 2 !!!

[honest rivet]

i tried with my Playstation controller and it worked xD so it has to be an issue with the input

[bronze maple]

wtf 😂

[honest rivet]

idk, sir^^ idk

[pastel oak]

yea as i said wokada might have troubles identifying audio interfaces like yours

[tepid knot]

so i have a promlem

i cant use voice changer

i have amd graphics card and it looks like this

i cant send screenshot

[graceful marten]

I was trying To Download the ai model in the cloud but I'm getting lot of errors , can anyone help me ?

[rare goblet]

Ayo? @graceful marten level 1 !!!

[pastel oak]

can you explain what you mean. does it say gpu: cpu ?

[tepid knot]

can we go dms?

[rare goblet]

Ayo? @tepid knot level 1 !!!

[pastel oak]

just ask in here

you can post screenshots now you got lvl 1

[tepid knot]

thats how it looks

and when i press start it just gave me some loud noises

[pastel oak]

can you first tell me the full name of your gpu

[tepid knot]

Radeon RX6750 XT

[pastel oak]

alright im gonna tell you which settings to use

F0 det: rmvpe_onnx
Chunk: 128 chunk
Extra: 8k (maybe 16k later after everything works)

Do you have virtual cable installed? If yes, select OUTPUT for your virtual cable. Your input and monitor are correct

Are you 100% sure that gpu1 is your normal gpu selection? If you are unsure, open task manager, go to performances, there on the left bottom side somewhere it will say GPU0 or GPU1

@tepid knot

[tepid knot]

it says gpu 0

[pastel oak]

then select gpu0

[tepid knot]

okay

[pastel oak]

you selected gpu1 on the screenshot i think thats why it didnt work probably

[tepid knot]

thanks

wait im gonna reboot and give feedback

i have vb installed

[rare goblet]

Ayo? @tepid knot level 2 !!!

[tepid knot]

but

it doesnt show in voice changer

ohh nvm

i restarted and it showed up

but

it works terrible

because im polish

and it trying to use japanes

do you know any polish female voice model?

[tight terrace]

what happens if the voice conversion breaks and goes on forever

[pastel oak]

I dont know polish voice models no

So you can technicslly try to upload index of a custom voice model and set it to a very low value like 0.2 and see if it works.

Id avoid using index but if you have a heavy accent it COULD help, but it might also just sound like autotune
It will use more resources though

[ionic venture]

it keep showing this every time i open the app

[pastel oak]

when you open the app or when you do something specific like selecting voice model or mic etc

• send pic of the console when it crashes

[ionic venture]

no i just download it a few min now

first it run normaly but few sec after that the error show up

[pastel oak]

well you must be clicking something for it to happen

[ionic venture]

after i click start

[pastel oak]

Click start then show what cmd shows

the Black background with white text

[ionic venture]

no im not see any cmd

it just show cmd bootsing the app when i click this shortcut that all

[rare goblet]

Ayo? @ionic venture level 2 !!!

[pastel oak]

Maybe something failed on installation

Delete the folders pretrain and model_dir then start the app

Also delete "stored_settings.json"

[ionic venture]

this is the thirt time i dowload the app because after i use it the next day i open it up like usual and it just show me white screen windows

and can't do nothing

[proper shale]

check command prompt

[pastel oak]

his command prompt closes instantly when it crashes

[rare goblet]

Ayo? @pastel oak level 30 !!!

[pastel oak]

alr asked for it

[proper shale]

hm

[pastel oak]

#🔍│help-w-okada message

try this maybe it wont close after crashing

[ionic venture]

no it close instantly when it open

[pastel oak]

ok do the thing i linked

[proper shale]

yeah then check what shad linked

[pastel oak]

It wont close

[proper shale]

might work

[ionic venture]

ok

[pastel oak]

Yea its weird that his command closes but the gui still runs thats the weird thing lol

[pastel oak]

Do you have an antivirus active? It might shut it off

Never seen this before

[ionic venture]

windows security ?

[marble seal]

how to use the .ckpt file? what app do i have to download?

[odd shale]

What did you try to download?