✨│ai-help

[terse siren]

nono i was on the site and i didnt see the big thing that said mainline

[low shard]

Ohh alr

[brittle wing]

hey guys i need some help

i have followed the instructions from realtime voice changing guide and when i go to the git link in the description, im not able to find a table with mac/windows (no table at all)

when i directly got to the hugging face link from there i couldnt find the cpu,cuda for windows link

please help me with this

[low shard]

whats ur pc gpu?

[knotty moth]

get 6 GB one, or better 3060 12 GB

[low shard]

i mean hes doing it on cloud so

[brittle wing]

oh oh got it running now

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

my bad downloaded the ml one instead of onyxgpu thing

[brittle wing]

3050ti 4gbVRAM

im going for kratosOG model

[low shard]

kinda bad but should work

did u fix the issue?

[brittle wing]

yes yes

thanks for responding 🙂

[low shard]

Oh alr then

[terse siren]

do i really seem like a male everyone keeps thinking im a dude when i talk to them 😭

[rare goblet]

Ayo? @terse siren level 2 !!!

[terse siren]

yeah let me just pull like 700 dollars out my ass rq

[low shard]

oh my bad, i didn’t see your pronouns so didn’t know your a girl

[terse siren]

yeah nah ur fine

[low shard]

didn’t mean to misgender or anything ofc

[terse siren]

i really dont care LOL

[low shard]

just saying 😭

[knotty moth]

I didn't mean the 4070/super/Ti, go check pricing in your region

[knotty moth]

I'd recommend fork wokada: https://rentry.co/forkvoicechangerguide

[low shard]

I mean GPUs are expensive

[knotty moth]

real

we couldn't expect 3060 drop below $200, that's it

[wicked ice]

does anyone know why that whenever i try to use the voice changer or just talk, i cannot hear anything come out?

[latent kettle]

Hello everyone. Can someone tell me how to train models on kaggle ?

[odd shale]

There you got a guide for kaggle. (Read the Cauthess one)

-kaggle

[azure marsh]

📘 Kaggle Notebooks

• Applio Notebook, by Vidal Kaggle
• Applio Notebook, by Shirou Kaggle
• Music Source Separation, by Shirou Kaggle
• UVR5 NO UI, by Eddy Kaggle
• RVC Mainline, by Hina Kaggle
• Original W-Okada's Voice Changer, Kaggle
• Modified W-Okada's Voice Changer, Kaggle
• 🆕 UVR5 UI, by Eddy & ArisDev Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline Kaggle by Cauthess

Note: Kaggle limits GPU usage to 30 hours per week.

[odd shale]

Have you tried reading the W-Okada docs?

[wicked ice]

Yeah

[rare goblet]

Ayo? @wicked ice level 1 !!!

[odd shale]

-rt

[azure marsh]

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

[nimble dagger]

@crude bolt come pls

help me ))0

[pastel oak]

check if theres any errors or pipeline messages in console

[azure marsh]

Suggestion for @brittle wing

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[stray dragon]

voice isnt changing

on any models

[odd shale]

What do you mean?

[pliant crest]

No CUDA GPUs are available

[low shard]

are you on google colab?

like be more specific about your issue

[pliant crest]

yep

It just says that

idk shit about the code

and on gradio itself too

[low shard]

what colab are u using (link) and can u send a screenshot of the error

be sure to not be following yt tuts, 90% of the time they are old

[pliant crest]

AICoverGen-WebUI, by SociallyIneptWeeb (modified by Hina)

im using that

[low shard]

are you sure you are on a T4 GPU Runtime?

there should be a 'T4' at top right

[jade raft]

WHEN I OPEN IT I ONLY GET THIS AND SOMEONE NEVER TURNS ON TO TELL ME WHAT TO DO

[low shard]

Like:

If there isn't, do runtime -> change runtime type -> T4 GPU -> Save
And re run from start

[low shard]

!howtoask

[patent trellis]

How To Troubleshoot

__**GIVE CONTEXT.**__ 📝

• Don't simply mention your issue, like "my rvc is not working".
• Describe the step you are on, what you're trying to do, the RVC you're using, a screenshot, etc.
• The more context, the better.

__**BE POLITE.**__ <:matsuripray:1159685390156967936>

• Don't be desperate. You can ping a Helper, but if they ignore, they aren't available/don't know the answer.
• It's okay if you're frustrated, but don't take it into this server.
• Don't DM without prior consent.

__**BE PRODUCTIVE.**__ 🤝

• Don't ask for every little instruction. Put your own effort & test things by yourself.
• Don't ask to ask.
• Check if your answer is a Google search away/on our guides website.

If you've changed your mind about your message, feel free to delete/edit it.

[pliant crest]

No

[low shard]

You need a GPU runtime to run this, do #✨│ai-help message

[pliant crest]

What colab do I use?

[low shard]

You can use that one

the problem is that you are on CPU

not gpu

follow what i said in the message

[pliant crest]

Im using that one

its still giving that error

[low shard]

could you show a screenshot?

because you just said u don't see the T4

[pliant crest]

bout what?

[low shard]

the google colab

[low shard]

alr it seems like you are connected, did you run in this order:

1. Clone repository
2. Install requirements
3. Run WebUI

[pliant crest]

Idk what any of that shit is

[low shard]

I will run it rq to see if there's an issue inside the colab code

[pliant crest]

the 2023 model worked

[low shard]

its the name of the code cells

[pliant crest]

wtf is a repository

[low shard]

You need to click the Play button to run them

[pliant crest]

and what is WebUI

[low shard]

repository = like a project shortly said

[low shard]

web (runs on the browser) ui (interface)

[pliant crest]

how do I run it?

[rare goblet]

Ayo? @pliant crest level 2 !!!

[low shard]

I just told you to run the cells with those names in the right order

[low shard]

#✨│ai-help message

You need to click the play button of each of those in the right order

@pliant crest i just tested it rn and it works

If you really cant understand it, want me to give you a written guide with images of another project to make ai covers ?

[pliant crest]

Thank you

Its working now

[low shard]

alright

your welcome

[inland grove]

Installed TensorBoard from the bat file on https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/, but it just opens to a blank white screen on the localhost. Does anyone know why this could be happening?

[jovial bluff]

is 70 seconds/% a normal install time

[ionic coyote]

How do you convert a model to ONX?

[rare goblet]

Ayo? @ionic coyote level 1 !!!

[knotty moth]

in RVC or wokada?

[ionic coyote]

i already find out! Thank you!

[wraith pine]

Hello, today I tried to download some rvc model, after I add it in rvc client, and tried to use my gpu, client just crash, but whenever i try to use cpu - it workes , that is only happening with downloaded rvc model, default works fine on gpu
https://i.imgur.com/85EQO9P.png

[pastel oak]

You downloaded AMD version

the "dml" on the top right indicates amd

[wraith pine]

very weird, cause like default rvc models whenever I switch on gpu is working, and yesterday all worked fine, but thanks ! I will check for nvidia version

[pastel oak]

Default ones are in onnx that could be a reason

either way should use the nvidia ones in any case ye names should have std in it for nvidia

[wraith pine]

and how I can choose what version I am downloading ? like there is only 1 file and I cannot chose which one

[rare goblet]

Ayo? @wraith pine level 1 !!!

[wraith pine]

I am downloading bottom one

[pastel oak]

i dont understand what you mean with this but i had it wrong, std is amd, cuda is nvidia mb

[wraith pine]

you said I have amd version of app, but how I can download one for nvidia?

[pastel oak]

cuda

3rd of the bottom

[wraith pine]

ohhh i see now

sorry

[odd shale]

-realtime

[azure marsh]

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

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[inland grove]

^^^

[worn lake]

hello guys

i need help

how i can make a voice model ?

[pastel oak]

https://docs.aihub.wtf/essentials/how-to-make-voice-models/

[shrewd kraken]

hello i need someone how can run rvc gui for me because i cant found how i can install all this stuff but i need to do something for the work i just need 5 minute of your time pls

[pastel oak]

You can use ilaria rvc zero no need to install anything

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[frank elbow]

GUYS WHEN I RUN THE APP IT TELLS ME" wait web server ..............10

what do i do??

[hushed steppe]

I wanna use medleyvox but is there a way to use it that isn't like downloading it? Is there a colab?

[rare goblet]

Ayo? @hushed steppe level 5 !!!

[hushed steppe]

Since I'm on android

[peak copper]

how to fix this error "UnboundLocalError: local variable 'embedder' referenced before assignment"

[low shard]

Medleyvox?

[hushed steppe]

Ye

It's to separate two audios

[low shard]

A voice model?

[hushed steppe]

Like if a singer is singing on top of another

It'll give you two tracks and filter them to not be on top?

Basically

[low shard]

People use UVR for separating those usually with the HP Karaoke 6 model
Idrk much about medleyvox

[low shard]

What are u using

[hushed steppe]

Medley vox is better than that one aaa

I'll see

[low shard]

From what I seen the latest update on GitHub is a year ago

Since your on android I could suggest you to use that model on a cloud uvr, but idrk about medleyvox sorry

[hushed steppe]

What's is cloud uvr?

[low shard]

Well, there are cloud ports of uvr (cloud is remote good pc, so running it on a cloud computing service like colab or hf spaces)

[hushed steppe]

I'm on a phone so

[low shard]

Or use the site mvsep which has uvr models but u got a queue

[low shard]

So yea u should use cloud

[hushed steppe]

I use mvsep already

[hushed steppe]

I have no idea how to go about that

[low shard]

https://huggingface.co/spaces/TheStinger/UVR5_UI or https://docs.aihub.wtf/rvc/resources/vocal-isolation/#cloud-uvr

[deep pagoda]

How can i create ai cover?

[rare goblet]

Ayo? @deep pagoda level 1 !!!

[low shard]

What's ur PC GPU?

[brave garnet]

RVC Colabs and Spaces

⠀

Google Colabs

⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.

AICoverGen-NoWebUI
Useful for making covers, doesn't include a UI, by Ardha, by Eddy, Hina and Gdr.

RVC Disconnected
To train new voice models, by Kit Lemonfoot.

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[sick wraith]

man if i don't quit getting these pydantic errors, i'm a scream

[brave garnet]

RVC Colabs and Spaces

⠀

Local Forks 🖥️

⠀
Mainline RVC
Original project, suggested for advanced users,
by the RVC-Project team.

Applio
Simplified, suggested for all, by the Applio team.

RVC Studio
Simplified, suggested for all, by SayanoAI.

Mangio-RVC
Simplified, may not be supported anymore, by Mangio621.

AICoverGen
Simple yet great way to make covers, by SociallyIneptWeeb.

Replay
From the greators of weights.gg, excellent product for everyone.
⠀

Made with love by Ilaria 💖

[dusky island]

So how do you know how many epochs are required for your model? is there some sort of math formula or something like that? I really want to get this right

[pastel oak]

Tensorboard graphs

[dusky island]

yes the total graph right? i've tried finding tutorials and guides in this server to help me better understand the tensorboard. but i still don't quite understand how im to make up the amount of epochs to train a model on, it kinda stayed steady at 932 steps but does that equal epochs or am i misreading the graph?

I have to start a new training session right now because my older one had used up its GPU for that account, so i will have to switch to an alt account and start the training again.

this was the graph

[pastel oak]

Not done training yet

https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/

[dusky island]

Given that you have quite the experience, I have a 15 min ADR interview recording of an actor, what would be the proper batch size for something like that, because i did find the guide on batch size but they refer to samples which i guess is the amount of .wav files the code has split the audio file into multiple parts. But how do you know how many samples you have?

im quite sorry for being a newbie. and probably having very easy questions

[pastel oak]

Assuming your gpu has ~8gb vram Id say use 8 batch size

Its more so about total amount of samples (15 mins), batch size determines how much of the audio is trained each cycle

[brave garnet]

RVC Colabs and Spaces

⠀

Google Colabs

⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.

AICoverGen-NoWebUI
Useful for making covers, doesn't include a UI, by Ardha, by Eddy, Hina and Gdr.

RVC Disconnected
To train new voice models, by Kit Lemonfoot.

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[fossil arch]

Is there anything for making an ai cover, and not a whole voice changer?

[pastel oak]

https://docs.aihub.wtf/essentials/how-to-make-ai-cover/

Local download RVC Mainline

Do you have a good gpu

[fossil arch]

one second.

[rare goblet]

Ayo? @fossil arch level 1 !!!

[brave garnet]

RVC Colabs and Spaces

⠀

Google Colabs

⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.

AICoverGen-NoWebUI
Useful for making covers, doesn't inclued a UI, by Ardha, by Eddy, Hina and Gdr.

RVC Disconnected
To train new voice models, by Kit Lemonfoot.

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[fossil arch]

I downloaded this one.

[fossil arch]

Unsure. I have an RX 580 (The AMD one, and yes I downloaded the AMD/Intel version).

[pastel oak]

Those are colabs, needed if your gpu is not good or you dont have one

Rx 580 is good enough for inferencing

As long as you downloaded RVC1006AMD_Intel you good. Follow the rvc "Inference" steps

Inference is the technical term for ai cover here

[fossil arch]

Thanks for letting me know.

@pastel oak I didn't know you guys had an API website, I downloaded the local version straight from the bot.

However, I think I made a mistake or something.

because I cannot find "go-web.bat"

[pastel oak]

Ahh yea the prepackage of amd doesnt have go web i keep forgetting

let me see if i have it on phone

[fossil arch]

I am training a voice, and it seems to be using my CPU, and not GPU, is this normal, (I have set the GPU index to 0 on all fields).

[stray dragon]

voice ai isnt launching

it shows a pop up and disappears

[fossil arch]

almost burnt my cpu

[pastel oak]

W8 you were asking for ai cover, not training a model. You cant train with AMD gpu the normal way

[fossil arch]

I want to make my own AI covers

[stray dragon]

Pipeline has been deleted
[Voice Changer] generate new embedder. (anyway)
[Voice Changer] Loading index...
[Voice Changer] Index file is not found

[rare goblet]

Ayo? @stray dragon level 1 !!!

[stray dragon]

my voice isnt being changed

[pastel oak]

Making ai covers and training a voice model are 2 separate things

For training youll have to use online, ill copy paste options in a sec

[pastel oak]

Delete the folders pretrain and model_dir then try again

[stray dragon]

uh i tried drifferent models and the preinstalled ones

none of them work

[pastel oak]

Training models per cloud

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
  • RVCDISCONNECTED (no ui)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
• Mainline (UI)
• Applio by Vidal (UI)
• Applio by Shirou (UI, no guide as of right now)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
• Mainline (UI, No guide as of right now)
• Applio (UI, No guide as of right now)

I recommend kaggle mainline

[pastel oak]

Ok? Did you do what i wrote

[stray dragon]

im sorry idk what that is

[pastel oak]

Those are folder names

XD

[stray dragon]

oh ok

[pastel oak]

Wait youre using voice.ai i overread it

We dont offer support for it

But the console you wrote seems like wokada idk

[stray dragon]

im not using voice.ai

[pastel oak]

Then delete those folders i wrote

Try again

[stray dragon]

alr

[hollow valley]

Hey everyone, I’m having an issue with RVC during vocal conversion. I trained a model on a rapper’s voice with a delicate, higher tone, but I’m trying to convert the vocals of another rapper who has a raspy, low voice. Even though I cleaned up the vocals and removed noise before conversion, I can still hear the raspy, low tone of the original voice coming through after the conversion. I’ve tried various parameter settings, but I haven’t been able to get a clean change in tone.

At first, I thought RVC mainly worked by swapping syllables using AI, but I realized that the voice needs to be more similar to the original for a better result. Even though both rappers have a similar flow, the sound quality isn’t what I expected – the raspiness from the original voice still comes through. I tested it with another rapper who has a slightly higher voice, and the result was better, but I’m still not sure if I’m doing something wrong with the settings to completely get rid of the raspy voice.

Can anyone help me figure out how to solve this issue?

[stray dragon]

same error

i tried both versions

one that supports nvidia graphics and intel one

both have the same error

[pastel oak]

Are you using alpha version

[stray dragon]

nope

5.3 i think?

[pastel oak]

Old version

Download .18a

https://rentry.co/voicechangerguide

[stray dragon]

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Tue_Feb_27_16:28:36_Pacific_Standard_Time_2024
Cuda compilation tools, release 12.4, V12.4.99
Build cuda_12.4.r12.4/compiler.33961263_0

cuda is working right?

[pastel oak]

Probably yea

[stray dragon]

[Voice Changer] generate new embedder. (no embedder)
[Voice Changer] Loading index...
[Voice Changer] Index file is not found
[Voice Changer] update configuration: passThrough false

still the same thing

[silver fulcrum]

might be a dumb question but does my pth file need to be 60 megabytes

[pastel oak]

Do you have antivirus

[stray dragon]

nope

[rare goblet]

Ayo? @stray dragon level 2 !!!

[pastel oak]

Yes thats just how it is lol

[stray dragon]

i have a very clean windows install

i have defender disabled too indeed

[pastel oak]

But this doesnt seem to be any errors

[stray dragon]

idk

i speak

and my voice isnt changed

ive looked at other youtubers using

[pastel oak]

Does it output your normal voice

[stray dragon]

ttheir console doesnt say this

[stray dragon]

yeah

[pastel oak]

Is your passthru red or green

[stray dragon]

red

[pastel oak]

XD

[stray dragon]

blinking red

[pastel oak]

Disable it

[stray dragon]

oh

its supposed to be off?

[pastel oak]

Yea

Red is normal voice

[stray dragon]

and how do i hear myself then

[pastel oak]

Monitor: headphones

[stray dragon]

BRO ISTG IF THATS THE ISSUE ILL KMS

😭

[stray dragon]

lets see

i do have my monitor device as my headphones

i dont hear anything

[pastel oak]

Id kms anyday of the week

[stray dragon]

thats why i turned on passthrough

i thought it meant "hear myself"

[pastel oak]

Can you check for errors while passthru is green

[stray dragon]

passthru is never green

its either white or red when i turn on

[pastel oak]

White?

Tf

Screenshot

[stray dragon]

wait leme show

can i even upload screenshots?

[silver fulcrum]

ah right thanks

[stray dragon]

green

ig

[pastel oak]

Counts as green

[stray dragon]

AH WHAT THE FUCK

[pastel oak]

The first pic wont load for me

[stray dragon]

IT WORKS

I SOUND ROBOTIC 😭

[pastel oak]

Send full screenshot of the app

[pastel oak]

Whats your gpu

[stray dragon]

1060

[pastel oak]

Youre using the wrong version

Lol

You got amd version

[stray dragon]

oh

[pastel oak]

https://rentry.co/VoiceChangerGuide

[stray dragon]

nah i just installed every one of them

[pastel oak]

Download nvidia

[stray dragon]

to see if which works

[stray dragon]

i have

leme open that

[pastel oak]

For 1060 id actually recommend the fork which is a diff program made by a developer here, has better performance for older cards, you can give it a try later if wanted

https://rentry.co/ForkVoiceChangerGuide

[stray dragon]

alr

does the same job though right?

[pastel oak]

Ye

[stray dragon]

uh

new error

[pastel oak]

You dont have gpu driver installed?

[stray dragon]

i do

i installed cuda drivers yseterday

12.4

[pastel oak]

https://developer.nvidia.com/cuda-downloads

Can you get this too

[stray dragon]

12.6?

[rare goblet]

Ayo? @stray dragon level 3 !!!

[pastel oak]

I didnt check what version it gives you

Probably

[stray dragon]

it gave me 12.6

[pastel oak]

Sure 12.6 then

[stray dragon]

doesnt that mean ill need to reinstall another version of pytorch then?

[pastel oak]

Nah

Voice changer uses prepackaged pytorch

[stray dragon]

okok

[pastel oak]

Python*

[stray dragon]

ive installed 12.6

should i try again?

same error :D

[pastel oak]

Download fork xd

[stray dragon]

you mean this?

bro what 😭

intel works???

shit it doesnt support uhd

do i need a 4090

[silver fulcrum]

ffs why is every pth 55mbs 😭

[pastel oak]

Somethings wrong with ur gpu

U got the drivers and toolkit you say?

[azure marsh]

💻 (Local) Realtime Voice Changer

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[dire minnow]

can someone pls send me the illaria link

[sick wraith]

When we download voice models, what's the best way to use them?

[odd shale]

You can use it either for conversion on RVC or realtime on W-Okada.

[sick wraith]

@odd shale you got the link for that w-okada? you referring to the voice changer that's written in all chinese?

[rare goblet]

Ayo? @sick wraith level 2 !!!

[tame zealot]

anyone know why my pytorchstreamreader is failing im assuming thats the reason my client doesnt want to work

[odd shale]

There you have.

-rt

[azure marsh]

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

[odd shale]

Read the Deiteris' fork guide.

[sick wraith]

@odd shale How do i install it in linux? i don't see instructions for it

[odd shale]

Deiteris fork got some improvements over the OG W-Okada

[odd shale]

Check this part on the guide.

[sick wraith]

ah ok, downloading the tar file now, ty

[odd shale]

You're welcome buddy.

[novel lotus]

imstuck on this dumb start_hhtp.bat thing it gives me all the text in the box and then doest do anything

[sick wraith]

@odd shale Got connection refused while trying to download the files

2024-10-03 21:39:18 (557 KB/s) - Connection closed at byte 1038090240. Retrying.

--2024-10-03 21:39:19-- (try: 2) https://objects.githubusercontent.com/github-production-release-asset-2e65be/741349909/753bfd42-1bb5-4dee-8936-685b0f9f7dc7?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=releaseassetproduction%2F20241004%2Fus-east-1%2Fs3%2Faws4_request&X-Amz-Date=20241004T010857Z&X-Amz-Expires=300&X-Amz-Signature=0bb6ede72e69ce68e8caf6e7483bb11d65f0c00cc716a8af2a709dbcbbd82836&X-Amz-SignedHeaders=host&response-content-disposition=attachment%3B filename%3Dvoice-changer-linux-amd64-cuda.tar.gz.ab&response-content-type=application%2Foctet-stream
Connecting to objects.githubusercontent.com (objects.githubusercontent.com)|185.199.109.133|:443... connected.
HTTP request sent, awaiting response... 401 Unauthorized

Username/Password Authentication Failed.
FINISHED --2024-10-03 21:39:19--
Total wall clock time: 57m 14s
Downloaded: 1 files, 2.0G in 57m 7s (598 KB/s)
Invalid number of options.

[odd shale]

No need to ping staffers.

But i don't know how to solve that error..

[sick wraith]

Ok

[gloomy nexus]

@odd shale i cant put in .wav files?

[wicked ice]

does anyone know why whenever i try to upload and use a custom voice it says this

[solemn grail]

does anyone have the latest link/app for using RVC models to make AI covers? I've been out of the AI cover game for a while and don't know what the latest stuff is. Even a link to tutorial would be a big help

[low shard]

What's ur PC GPU?

[low shard]

Ilaria RVC zero Is the latest one, which is a ZeroGPU (A100 paid by Ilaria) HuggingFace Space

(only for inference, which is using the models)

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[simple ore]

trying to use v2 model with v1 setup? or other way around

[wicked ice]

how do i download th v2 setup?

[simple ore]

what application are you trying to use?

[simple ore]

repo link

[wicked ice]

huh?

[simple ore]

where did you get it

[wicked ice]

https://drive.google.com/uc?id=1AWjDhW4w2Uljp1-9P8YUJBZsIlnhkJX2&export=download

there

[simple ore]

yeah, it is old v1 stuff

[wicked ice]

how do i get v2?

[simple ore]

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/releases/tag/2.2.231006

get the prebuilt nvidia or amd/intel

[wicked ice]

this?

[simple ore]

yeah

[wicked ice]

then what?

[simple ore]

then unzip

[wicked ice]

okay

[simple ore]

and there's realtime .bat

[wicked ice]

whats that?

[rare goblet]

Ayo? @wicked ice level 2 !!!

[simple ore]

"go-realtime-gui.bat"

[wicked ice]

okay

[wicked ice]

ive opened it now what?

[stray dragon]

yes

[pastel oak]

@wispy lodge sorry for ping as always xd
Do you know what else could cause this? 1060, got drivers and toolkit installed

[stray dragon]

buy me a 4090 pls

[knotty moth]

get a job

[stray dragon]

im too young

[stray dragon]

idk if that would be enough to easily afford a 4090

[wispy lodge]

This doesn't seem to be a fork version. Show full log pls

Fork should be ignoring installed cuda toolkit and it MUST NOT be installed because you may end up with incompatible cuda toolkit version that doesn't work like in this case

[solemn grail]

I actually use a MacBook

[rare goblet]

Ayo? @solemn grail level 1 !!!

[scarlet musk]

HOw can i make my own voice modes

model

[mint horizon]

so im using rvc, im wondering if theres anyone who knows a good microphone for okada rvc

would anyone be down to let me know what microphone theyre using and their experiences?

[jaunty shale]

just a quick question. I am training in RVC Disconnected with a 12 minutes dataset. What batch size and pretrain type is mostly recommended for this?

[simple ore]

4 and default pretrain

[stray dragon]

what do i do now

[wispy lodge]

Show full log with error

[sick wraith]

I got the program installed, how do i upload a voice?

[low shard]

oh, you can only inference (use models) on macbook locally (on ur pc), the guide would be https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md

Or, if you want an easier way, Ilaria RVC Zero is the fastest cloud option for inference

[brave garnet]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[low shard]

whats ur pc gpu

[solemn grail]

Thank you!!! 💙

[solemn grail]

💙💙💙

[wispy lodge]

1. This is not a fork version.
2. Delete CUDA toolkit.

[stray dragon]

oh wiat

i forgot to use fork

sorry

same type of error on fork too

RuntimeError: Error in execution: Non-zero status code returned while running Pad node. Name:'/Pad' Status Message: CUDA error cudaErrorNoKernelImageForDevice:no kernel image is available for execution on the device

[wispy lodge]

Could you check nvidia driver version just in case?

[rare goblet]

Ayo? @stray dragon level 4 !!!

[stray dragon]

there is an update

[wispy lodge]

And you say you have 1060? 🤔 Could you try downloading version b2300? Maybe latest update somehow affected 1000 series
https://huggingface.co/Shadicti/deiteris-Fork/blob/main/Windows-Nvidia-b2300.zip

[stray dragon]

1060 maxq to be exact

its the normal 1060 laptop gpu just more underpowered

for thin and light laptops

it doesnt support some features like nvenc and display output

could the maxq be causing a problem with cuda?

all nvidia gpus have cuda right

[brittle wing]

someone help the console on start_http isnt opening

[stray dragon]

i dont think its the version

[pastel oak]

#🔍│help-w-okada message

[stray dragon]

i get the cuda error everywhere

[brittle wing]

oops my bad

[wispy lodge]

They all have yeah, just capabilities vary. But from what I can see 1060 max q has the same capabilities as desktop 1060, so shouldn't be a problem

[stray dragon]

ig

[jovial dock]

guys

where the fuck do i get the classic voice changer

[stray dragon]

leme try uninstalling cuda toolkit

[pastel oak]

Whats got you so mad

https://rentry.co/VoiceChangerGuide

[stray dragon]

uh but idk how to 😃

[jovial dock]

thanks

[pastel oak]

Either the way you usually uninstall programs or might be a uninstall function if you start up the toolkit thing

[neat hatch]

[stray dragon]

you mean just uninstall the nsight things?

[pastel oak]

I dont know if its that but if it says installed today then probably

Id rather you go through the toolkit thing for an uninstall button

[stray dragon]

wat

im not understanding 😭

sorry

[stray dragon]

it got installed with the cuda drivers

[pastel oak]

Idk what you want me to say lmao

The place where you uninstall apps and programs, go in there, check if the toolkit shows up there

Or

Open the toolkit folder, there could be an uninstall button xd i never had to download toolkit so im rlly not sure

[stray dragon]

yes yes

found it

[pastel oak]

Type shi

[stray dragon]

you want me to just remove toolkits right?

[pastel oak]

Yas

[stray dragon]

yes

done

now what

do i install them again :D

i cant even use cpu 😭

[sick wraith]

i trained my models using xtts-finetune-webui, but it creates a model.pth and a json file, not an index file. Am i using the wrong type of model?

[pastel oak]

Is the screenshot the fork or original

Try the fork again then also try the b2300 version emoji linked

[stray dragon]

fork

[knotty moth]

• try different driver version that may be more stable
• clean up using DDU then reinstall the driver
• fresh install windows as the last resort

[sick wraith]

also is xtts or rvc better for voice cloninig?

[pastel oak]

theres usually a separate button for training index, not sure how xtts does it

we all use rvc here would recommend it yea lol

[sick wraith]

thanks, also what metrics can i study to see if the training is good or go to crap?

i see loss_gen loss_fm, but would something like a tensorboard be better to use? i haven't really used them much in the past. I think i've set up like 1 before

[pastel oak]

Tensorboard is generally used, yes

Bit of explanations can be found here:
https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/

[odd shale]

Actually you must care mostly about Loss_g_total and d_total.

And read the guide Shad gave you above.

[odd shale]

🐢

[pastel oak]

g/total is looked at most commonly, its the most "overall" if that makes sense

[pastel oak]

🥬

[odd shale]

🐢 ❤️

Hi Shady!!!

[pastel oak]

hi leooooo 🥺

[pastel oak]

If you want to learn about other important factors, you can refer to this message. But looking at g/total is enough for a high quality model in my and other peoples experiences.
If you like to be technical and learn your stuff, feel free to
#✨│ai-help message

[stray dragon]

i am not reinstalling windows

i recently did 😭

any other alternative to this voice changer?

voice.ai sucks

plus it was very laggy

[pastel oak]

https://rentry.co/rvcrealtimeguide

[knotty moth]

have you tried all but the last one I said?

[sick wraith]

any instructions for linux users? bat files don't work on linux. I can hunt around and see what i can find

[stray dragon]

wait no this is somethting else

[fallen wing]

Need some help in the past RVC was working fine but now when i play on games and use the voice changer in game for some reason it repeats what my teammates are saying??

[stray dragon]

another option is to buy a 4090 🤩

[sick wraith]

i see these instructions - tensorboard --logdir=path/to/your/logs are these the logs to the rvc project we are trianing at the time?

[rare goblet]

Ayo? @sick wraith level 3 !!!

[knotty moth]

a 2060 or 3060 wouldn't cost you a kidney

[odd shale]

🐢

[pastel oak]

No clue 😅

[stray dragon]

i sell kids

[pastel oak]

RVC has a folder thats simply called "logs", its where your voice model training info is saved

So youd link that folder, yes. And it will retrieve all info while training simultaneously

[sick wraith]

should i set the log path to the root log directory or the one i'm training on at that time?

i'll try both and see what happens

[fallen wing]

Uhhh need help why did it switch places?

Now its not even working

Theyre not showing up on what theyre supposed to be on

[dull crypt]

hey can someone please help me find the adam legacy tiktok voice for Free without having to pay for eleven labs

[fallen wing]

Need help why did it switch places?

[odd shale]

Adam Legacy voice?

Maybe you can look for tutorials on Youtube.

[fallen wing]

It was fine an hour ago but now it switched same as the selections are switched

[brittle wing]

I need to check some stuff when it installed python, is there a way to uninstall this voice mod and it's additional files that it unpacked? i see no .bat or .exe file to run the uninstall, do i just delete the entire thing?

[pastel oak]

Just.. switch back? Lol
It can randomly change spots

You can safely change it

[fallen wing]

Uhhhhh it stuck like that even after countless refreshes

[pastel oak]

I dont understand, the way you set it up first time you can change input and output again though no

[fallen wing]

Yes i can change it but its setting itself to what is supposed to be the output for input, and input for output

[pastel oak]

The screenshot is correct

[fallen wing]

Oh wait really am i high?

[pastel oak]

Dont get confused by vb cables terrible labelling

Ya its just the labels are inverted

[fallen wing]

Ight trust then

Yeah it works

What the hell

Labellings bruh

[pastel oak]

The label is so users know the input is where they first use it

But like

It creates more confusion

[sick wraith]

so does overtraiing commence typically when the los value starts to rise up? sounds almost seems like i could have possibly kept training here and gotten it even better

also how do i equate steps to epochs to know which epoch to choose?

[pastel oak]

Yea that models not done yet

Once it starts rising up its a potential OT (it could still fall down again later)

[sick wraith]

so as long as it goes up and keeps going up then itos OT

[pastel oak]

If it goes up, the last lowest point is the best version of the model, yes

[sick wraith]

i'm assuming the bigger the audio file or data set, the more epochs you need to train?

generally about how long of audio should we use? or is it dependent on the dataset

sorry i have so many questions, i'm a very inquiisitive person

[sick wraith]

looks like total steps / total epochs is my answer

[simple ore]

with a large data set (1hr+) the model has more information to extract

so as long as data set is not 5 hours of repeated "hello"

with more data the model reaches lower loss values faster than with less data

as the model shuffles audio samlples for every epoch it has more combination to go thru, so more epochs total may be warranted

[sick wraith]

it's like a 10 minute file, does the training work better with 1 big file, or many smaller files? or doe sit matter?

[simple ore]

one big file without slicing it? that's not a great idea

for better results perhaps slice it yourself

here's batch 4 tests, blue is 1 minute sample, others are 10 minute

the model got pretty much everything it could by 200 step and after it is unlikely there will be any improvement

[proper shale]

Interesting

[sick wraith]

can we use gpt-sovits to slice the files i'm assuming

yeah cause i'm up to 300 epochs still now and it's still trying to learn it

[proper shale]

I don't think slicing is necessary tbh, but if you want to do that sure

I think you can do that with Audacity and labels as well

[simple ore]

if you only have one big file the model has no choice by trying to emulate just how that sounds

so it may get good at doing that, but really bad at anything else

[proper shale]

In the end preprocessing automatically slices it up every 3 seconds so it shouldn't make a difference... at least in theory

[simple ore]

training with sliced audios re-arranges the sets randomply, so it learns to imitate a large variery of audio sequence

default 3 sec preprocessing sometimes slices audio on a middle of a word, that's not good

but does not really matter in the big picture

[proper shale]

well yeah that's what RVC does

for the funny

[simple ore]

of course it would be better to slice it manually

[proper shale]

[sick wraith]

you guys have been incredible so far, ty

[analog obsidian]

do you think it would make an audible difference slicing it manually?

[simple ore]

probably not 🙂

slicing is probably one of the lesser priorities. Clean audio > amount of sample data > batch size > slicing

[sick wraith]

so is bigger drops better when training, or is it better when it stays kind of flat

like mine before with the 1 audio file, it had a bunch of dips, this training set has been relatively flat

[simple ore]

another example

orange - 10 min, light blue - 5 min audio, dark blue - 1 minute audio --- all of those with a default pretrain, white - 10 min with Titan pretrain, 32k, batch 4

1 min audio does not have enough to shape the pretrain into properly reproducing the inferred audio

5 min audio does have a decent amount, so the output is okay

10 minute audio produces pretty good result

all comparisons are done using models saved at the same ~750 steps

[sick wraith]

so you're saying i should use a 10 minute file over split mini samples

[simple ore]

but a 45 min sample set sounds crisp af

[wild vale]

How does one make this useable when playing video games

When I play games I tend to just robot

[sick wraith]

all i'm asking is should i use one big file or use something like gpt sovits to slice them

[wild vale]

And the voice cuts in and out.

[sick wraith]

or are you saying it depends

[simple ore]

use slicer, dont use one big file

[sick wraith]

got it

[simple ore]

if you can cut it manually into 2-5 sec chunks so the cut is done at silence, do that

[sick wraith]

yeah cause i'm about to 5.6k steps now and the line is just flat, does that mean it's still learning?

[wild vale]

What settings do I use

[simple ore]

technically it is never flat, the loss varies for each step, it is just the variation can increase and at some point starts to increase

as long as kl, mel charts keep going down you can keep training, but it will be a miniscule improvement

[wild vale]

Is there someone who can help with rvc?

[rare goblet]

Ayo? @wild vale level 1 !!!

[rare goblet]

Ayo? @simple ore level 6 !!!

[simple ore]

this 5 min set is more or less done at 1.2k steps

[simple ore]

you can try giving python higher priority in task manager

and lowering the game's priority

[wild vale]

That messes with the changer

the issue is the buffering increases immensely

I have to tab out just for it to go back to normal

[simple ore]

yeah, there's no prioritization for gpu work

trying to train a model or realtime inference and play some heavy 3d game is not a good idea

[sick wraith]

so pretty much i could stop where my red dot is at

[rare goblet]

Ayo? @sick wraith level 4 !!!

[wild vale]

that doesnt help

[simple ore]

check 'ignore outliers' and click

[sick wraith]

ah that is much different lol

so i was looking at a macro viewing so to speake

i noticed there is the solid line and the faded line, what's the faded line for?

[simple ore]

faded line is actual loss per step, solid is the smoothed

[sick wraith]

here if i'm zoomed in

400 epochs

[simple ore]

very small epochs, you only got 8k steps

[sick wraith]

i'm assuming that's based on batch size?

right now my batch size is set to 6

and split my 10 minute file into about 45 files

[simple ore]

8000/400 = 20 steps per epoch x 6 = 120 samples per epoch, ~ 6 min audio normally

[sick wraith]

is there something i should have done different with my training?

[sick wraith]

unfortunately the person i'm doing the technology was much older and so getting clear voice samples is difficult

is that arnold?

[simple ore]

no, a russian speaker with a funny speech defect

[sick wraith]

ah ok gotcha lol

[simple ore]

but Arnold does sound close

same kind of hard R and H

[wild vale]

I need someone who can help its not working anymore

Im testing the Line 1 cable and now discord isnt even detecting it

All I did was change the input and output settings and when I changed them back it doesnt detect them anymore

[dusky wind]

Question: Should i use the "[experiment].pth" or the "G_2333333.pth" to generate rvc audio? Why is the second one so much bigger in size? Will it produce a better result?

[sick wraith]

damn that's good

What index file are we wanting to copy from rvc webui over to voice changer?

the added?

[analog obsidian]

model.pth is your model in a smaller size*, use this
g.pth is your generator's model, same as model.pth but bigger size because contains stuff related to training, e500.pth will sound exactly the same as g500.pth

[dusky wind]

thanks!

[rare goblet]

Ayo? @dusky wind level 2 !!!

[dusky wind]

I thought only "D_2333333.pth" was used for training stuff

[sick wraith]

so the voice i'm doing is in german, how would i translate that over?

[dusky wind]

Your dataset with audios in a certain language will work for audio made in other languages

[wild vale]

Who here knows how to use rvc

Its being detected in discord but in chat it isnt being detected

Where is the support for this thing?

[sick wraith]

what program should i use for text to speec with rvc?

[low shard]

RVC is natively for Speech To Speech, but forks such as ilaria rvc mainline & applio have built in tts (using Microsoft Edge TTS to make a generated tts audio, which i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs
• While Ilaria RVC Mainline here (no guide as of right now)

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero (Running on A100 GPU, free fasted rvc on cloud) and the guide

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

[analog obsidian]

G and D are needed for continue training and also can be used for training as well tho the latter is for more experimented rvc users
if you don't want to train anymore you can delete those ( G & D) to save space

[sick wraith]

i have a good pc, but ihave 12gb of vram

[low shard]

However, if you want good TTS, I would suggest you using actual TTS Programs, such as fishspeech (if u care about speed, 0 show so needs just an audio file without explicit training), or GPT SO VITS (best quality, few shows which means it needs some training)

[vivid pewter]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[sick wraith]

ok so rvc for speech to speech, than can do xtts for text to speech it seems

i did train an xtts for text to speech and it turned out pretty well

although it would be nice to stick with one technology and not bounce back and forth cause that will get confusing

[low shard]

i mean u can use technically rvc for tts but its not as good as an actual tts program ye, for tts u could use xtts or gpt so vits which should be better

[vivid pewter]

-gui

[azure marsh]

Yeah, I still use RVC Easy GUI.

[wild vale]

What in the--- are my questions being ignored?

What the hell is going on.

[dusky wind]

your questions are not clear

[dusky wind]

i don't even know what this means

[lime condor]

why on beatrice is work on rvc no?

[sick wraith]

is there aw iki on this realtime voice changer client?

not sure if i should do client or server or what

and it keeps crashing and asking me to refresh the tab

ok refrresh tab issue was my browser

[low shard]

I'm not really a wokada helper, but from what i know from other helpers:
if only the beatrice model working, its a failed install, you need to delete pretrain folder and model_dir and let it reinstall those

[low shard]

normal version: https://rentry.co/VoiceChangerGuide
fork: https://rentry.co/ForkVoiceChangerGuide

those are the local guides for wokada

[sick wraith]

does thiss program work for live voice changing or only for saving voice samples to wav files?

[proper shale]

Both for live voice changing and audio conversion

So yes

[sick wraith]

i gotta f igure out howi'm going to get this to work with easy effects

[broken jolt]

please tell me what to do, for some reason my output shows my headphones instead of microphones. Reinstalled the but it didn't help.

[pastel oak]

output is supposed to be playback devices, whats the issue

a mic cant be an output

[sick wraith]

Anyone here running Arch Linux?

[tacit solstice]

how I set this up

[brave garnet]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs

F0 Det.: rmvpe (suggested for all series)

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

Advanced Settings

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

⠀

Made with love by Ilaria 💖

[vale osprey]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

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

[dull crypt]

does anyone know good free TTS on hugging face, illaria rvc has the option for TTS but it doens't have the voice that i want which is the Adam (legacy) tiktok voice that everyone usees from eleven labs.

[simple ore]

https://www.weights.gg/models/clm73135c1ddgcctcuwh80uvn

this voice?

[slate zodiac]

I have old audio is so badly degraded that inferencing often gets the words wrong. Is there a fork that has a feature similar to GPT-SoVITS text-for-reference-audio?

BTW, I tried using GPT-SoVITS just for that feature, but I couldn't really get too far because of the lack of documentation.

[dull crypt]

thank you.

btw is it possible to use a pth file and strictly from the pth file use TTS for example luffy from one piece you fine the pth file its very accurate but when you merge his voice so that he can say something it changes the voice. So would it be possible to JUST use the pth file and type in characters and the voice created is only that from the pth??

[fading tulip]

hello, the voice i get isn't good. what i can do to make it better?

[molten flare]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[distant turtle]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

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

[simple ore]

explain again what you want to do

[low shard]

Nope, you can't use tts natively on rvc

Rvc forks generate a tts audio and use it as an input

There are different Text To Speech (TTS) AIs:

GPT So Vits: RVC isn't as good as GPT So Vits for tts, but gpt so vits (few shot tts, which means needs just a lil training for models) can't use rvc models (and viceversa), and its only limited to: english, chinese & japanese, if you wanna check gpt so vits instead, read https://docs.aihub.wtf/tts/gpt-sovits/

Freemium 11labs: A easy way to do TTS is https://elevenlabs.io/, you can't use RVC model on this but its a mostly premium easy way for good quality TTS

FishSpeech: FishSpeech is a 0 shot (no explicit training needed) TTS, if you got a good pc you can use it locally else use their site

With RVC Models:

RVC is natively for Speech To Speech, but forks such as ilaria rvc mainline & applio have built in tts (using Microsoft Edge TTS to make a generated tts audio, which i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs
• While Ilaria RVC Mainline here (no guide as of right now)

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero (Running on A100 GPU, free fasted rvc on cloud) and the guide

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

It's better to use an actual tts program like fish speech or gpt so vits

[granite cedar]

Hello guys,

I am trying to use the space "The Stinger" by Ilaria RVS.

When I all used my credits, the say to sign up on hugging face. So i'v signed and paid the pro version.

But I stay enable to generate some new voices...

What I did wrong ??

[pallid field]

Hi, does anyone have any links to guides/videos or advice of their own on how to sound more natural when using RVC? So far I either sound robotic with low pitch, like a gnome with high pitch and index makes the voice sound less natural, with "tone" being "capped", if that makes sense

[simple ore]

if you're making your own voice, a good quality mic and the absense of noise is the key

use wind filter, dont blow up the mic

[pallid field]

Those are present, I haven't had a problem with outside noise so far

[rare goblet]

Ayo? @pallid field level 1 !!!

[pallid field]

Oh

My bad, you meant a model after my own voice

No, I was using the RCC client as a real-time voice changer

[simple ore]

ah, that's different 🙂

[pallid field]

Wrong channel, my bad

[warm storm]

anyone have any good voice models that really sound like a girl?

and what are the best settings to make the voice louder and remove background noise (like pc fan etc)?

[simple ore]

nVidia's broadcast app removes noise from mic input

[low shard]

It's called Ilaria RVC which is a fork (modified version) of RVC, the creator is Ilaria aka TheStinger. It's a ZeroGPU Hugging Face Space, ZeroGPU is one of their service that offers an A100 by the creator of the space who pays it, everyone using their space has a quota on the account which you cant check, if you sing up to pro you will have X5 more the quota, but still not unlimited

Also, that HuggingFace Space is ONLY Inference (using models) for training (making models) you need to use something else

[limpid trout]

Alright
I used The voice models for my work.
Now I wanna learn how to make your own voice models

[limpid trout]

Hmmm like what

[low shard]

I remember we talked before and you said you got a bad PC so you need to use cloud

As you dont got a good PC, its better you use cloud for training an RVC Voice Model:

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
  • RVCDISCONNECTED (no ui)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio by Vidal (UI)
  • Applio by Shirou (UI, no guide as of right now)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • Mainline (UI, No guide as of right now)
  • Applio (UI, No guide as of right now)

Google colab is the easiest but u could get disconnected losing ur work

Kaggle is harder and requires number but it gives u way more GPU time

[limpid trout]

Ok

[elfin quiver]

From here on, it's completed in 1 second and are not finalized. can anyone help me?

[latent kettle]

Why this happened

[simple ore]

run out of vram?

[latent kettle]

Idk

Just tell me what is happening

Have gtx 1650

@low shard

[simple ore]

4gb is tough

[latent kettle]

How do I use it on my CPU

@simple ore

[simple ore]

well, i can think of two methods

edit configs/config.py

[junior parcel]

stupid question but generally speaking does more epoches mean a smarter ai voice kinda deal or

[rare goblet]

Ayo? @junior parcel level 1 !!!

[simple ore]

find if torch.cuda.is_available(): and change it to if torch.cuda.is_available() and 1==2:

[low shard]

Dunno, I don't really do local much

The only thing i found about the error is this

[simple ore]

more epoch = how long the training gonna shuffle audio samples and trying to reproduce them

if you have a deck of 5 cards, no matter how much you shuffle there's no benefit over repeated attempts

[latent kettle]

To me ?

Where

[junior parcel]

Fair enough do you know where to find like pretty decent AI like voice models?

[simple ore]

configs/config.py

[low shard]

its an unit of measurement of the ai cycle of training, does not mean more quality. There isnt a right epoch, check https://docs.aihub.wtf/rvc/resources/epochs--tensorboard/ if u are going to train a model

[junior parcel]

Like really well trained sorts of ones

[junior parcel]

ahhh will do

that makes more sense

[low shard]

You can search rvc ai voice model at:

• #1175430844685484042
• #🔍│find-models
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.aihub.wtf/essentials/how-to-make-voice-models/

But the quality can vary

[simple ore]

other alternative is uninstalling cuda tools, so torch.cuda.is_available not longer returns true

[junior parcel]

Yeah thats what I mean I was wondering if theres a specific place to find really safisticated high quality ones

But I'll just scour till i found the right one

[latent kettle]

Just tell me how to do that

[simple ore]

I told you

[latent kettle]

Which file

[low shard]

really satisficated high quality ones no, those are all the sites where u can find rvc models, which usually are good quality

[latent kettle]

I have to edit

[latent kettle]

Name the file

[simple ore]

I named it twice

[low shard]

btw the not enough memory seems like a memory issue to me

[latent kettle]

Can you tell me the exact path

[simple ore]

for the third time

configs/config.py

under your unzipped RVC1006

third alternative it setting an environment variable so no cuda devices are available

CUDA_VISIBLE_DEVICES

set it to empty value

[latent kettle]

Where ? Same file ?

[simple ore]

" setting an environment variable "

[latent kettle]

So I have to add an environment variable

[latent kettle]

This

This is that variable?

[simple ore]

it is a name of it

[latent kettle]

Can you send me a short video what I have to do exactly. Just getting confused

[pastel oak]

Send a screenshot of your rvc ui before you hit convert

[simple ore]

after that start rvc again

[latent kettle]

Everything looks fine @pastel oak

[pastel oak]

Maybe your rvc cant access your Users folder

Can you put the audio somewhere else and try again

[simple ore]

you're suggesting silly things

as the error is clearly shown

it has nothing to do with the files, it did not even get to loading the file

[pastel oak]

Has worked before

[simple ore]

it tried to initialize rmvpe f0 extractor and failed because pytorch ran out of memory

[rare goblet]

Ayo? @simple ore level 9 !!!

[latent kettle]

Not working

@simple ore

[simple ore]

okay... how much regular memory do you have?

[latent kettle]

RAM 16GB
GPU 4GB

@simple ore WHAT TO DO NOW ?

Drive space 210 gbs

[simple ore]

well, looks like Python is running out of memory and crashing

[latent kettle]

How to fix that

Should I download rvc-beta.7z ?

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC-beta.7z

@simple ore please help

[simple ore]

regular inference should not take that much memory, maybe 2-3 gb

dont know whether mainline RVC does anything different

[latent kettle]

Why here is false ?

[simple ore]

that's correct, CPU is FP32

is_half is FP16 precision

[latent kettle]

What ?

[simple ore]

https://en.wikipedia.org/wiki/Half-precision_floating-point_format

[latent kettle]

If everything is fine why it is not working

[simple ore]

open task manager / performance

see the memory

run inference, see what happens

[latent kettle]

Is there any alternative?

To use RVC ?

Should I go with Applio ?

[simple ore]

applio from command line

[latent kettle]

Or should I go to RVC studio ?

[simple ore]

or rvc command line

[latent kettle]

It's difficult

[simple ore]

not really, just using a .bat file or command prompt

pass parameters for inference such as model and index name, input and output files

[latent kettle]

Should I install RVC using Command line

[simple ore]

not what I said

[latent kettle]

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC-beta.7z

What about this ?

[simple ore]

looks old af

[latent kettle]

Will it work

[simple ore]

it is older than the one you're using right now

[latent kettle]

Probably it will work then

[latent kettle]

If it not. I can never use rvc on my system ?

And please give me a guide to use Command line RVC

[fallen flint]

when i dwonload a file, how can i turn a zip file into a sound file? Just extracting it doesnt work

[elfin quiver]

up up

[brisk quarry]

false

@low shard explain it

[latent kettle]

Do you know what is the error?

[brisk quarry]

what

[low shard]

cpu usually uses fp32

[brisk quarry]

thats not what i meant

u were supposed to say dont use cpu

[low shard]

u said 'false' to 'cpu is fp32' 😭

[brisk quarry]

why doesd using split audio make it twice as long

[rare goblet]

Ayo? @brisk quarry level 31 !!!

[wild canopy]

Guys, any model does not want to work without an .npy file, is it the same with you?

[simple ore]

.npy is a feature saved as np array, spec.npy are spectrograms of samples

the voice model (weights) itself does not need them

[pastel oak]

Voice changer?

[wild canopy]

Guys, I used MMVC, and now I use VCC and everything works

[pastel oak]

You should be using type RVC for model uploading

Which version of wokada are you using

[wild canopy]

I used RVC in MMVC but it required .npy, then I decided to download VVC and it no longer required .npy and the errors in the console disappeared as a result

[rare goblet]

Ayo? @wild canopy level 1 !!!

[pastel oak]

Didnt answer the question but sounds like you used one of the outdated versions
If alpha vcc works cool

[wild canopy]

I just switched from MMVC to VVC alpha

[simple ore]

i dont think it requires much effort, really

[rare goblet]

Ayo? @simple ore level 10 !!!

[simple ore]

there's russian dude and a dude from New Jersey speaking english with their specific accents

[distant turtle]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

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

[amber crow]

which fork of wokada is the best rn?

[wild canopy]

hеlp

[simple ore]

well, make it shorter?

[golden walrus]

guys, can i ask why the tensor board not running correctly ? 😦

[rare goblet]

Ayo? @golden walrus level 4 !!!

[golden walrus]

i have sync graph on but it stuck here

[simple ore]

close and re-open

[simple ore]

dont open it before the actual normal training starts

otherwise it will be stuck on the sync part

[golden walrus]

ah i get it

let me try

thank you in advance xd

ah it works

got it xd

[small leaf]

nihat hatipoğlu

[snow needle]

Guys I'm new here how do I download rvc?

[azure marsh]

Hello, @snow needle! Here are some recommended resources for you!

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde

[snow needle]

it says connection not secured for AI HUB docs

[crude bolt]

Hm. you are correct

[snow needle]

I just wanna know how to download it

[rare goblet]

Ayo? @snow needle level 1 !!!

[crude bolt]

alright; do you have an nvidia graphics card in your pc ?

there are multiple setups to download so i need to know a bit of your computer before being able to guide you to the correct resource

[snow needle]

yeah I do

I have an Acer laptop

[crude bolt]

are you on windows 10/11 ?

[snow needle]

Actually wait

Yeah

[crude bolt]

brand i dont have to know 🙂

[snow needle]

I'm on a predator triton

I also installed python cause I think I needed it right?

Some youtuber said so

[crude bolt]

https://github.com/deiteris/voice-changer/releases

On this page you will see:
voice-changer-windows-amd64-cuda.zip.001
voice-changer-windows-amd64-cuda.zip.002

This is a multipart-zipfile. You need both parts to extract the first file.

download the 2 files i mentioned.

[snow needle]

Downloading

Second file is almost down

*done

[crude bolt]

you will need both to extract the first file.

[snow needle]

Which app do I use to open these?

[crude bolt]

go in directory called MMVCServerSIO, then run MMVCServerSIO.exe

then wait 5-10 minutes.

it downloads alot stuff.

this is only like first time - the program checks weather you got all the files you need , if not it downloads.

[unkempt basin]

can I have help on why the custom voices I added, only slightly change in the pitch and nothing else

[rare goblet]

Ayo? @unkempt basin level 1 !!!

[unkempt basin]

while a preset model doesnt have the problem

[crude bolt]

do you have a nvidia card ?

[unkempt basin]

I tried justin beiber as a test and its just me with a different pitch

yes

but a random one is strong

[crude bolt]

hm interesting - i assumed you to say no here ... can you show me in the voicechannels.

[unkempt basin]

idk if it works on discord

ive not even gotten that far

[crude bolt]

no worries - if you got discord app you can stream the application

[brittle wing]

2 days and no response, maybe no one knows? I need to check some stuff when it installed python, is there a way to uninstall this voice mod and it's additional files that it unpacked? i see no .bat or .exe file to run the uninstall, do i just delete the entire thing?

[simple ore]

depends on how you've installed it

if it was just unzipping one big file into a folder, then just delete the folder

if you have a virtual environment, activated it, then cloned a repo and ran pip install manually, just delete the folder

if you just cloned a repo and ran pip install manually then all the packages are in the global... rip

[rare goblet]

Ayo? @feral marsh level 3 !!!

[brittle wing]

I installed the entire .zip then extracted it into a different folder

[simple ore]

if you mean downloaded a big .zip... then as I said 1st option, just delete the folder

[brittle wing]

Ah will it need to install stuff from .bat again?

[simple ore]

depends on what .bat you're talking about and what RVC package

[brittle wing]

@crude bolt https://huggingface.co/wok000/vcclient000/tree/main

[feral marsh]

why is rvc's delay super long all of a sudden? it takes like 3 seconds for the voice to process when before it was like 1 second

[simple ore]

@brittle wing if that's the application you used, just delete the folder

[simple ore]

is it still running on GPU?

[feral marsh]

yess

i didnt mess with anything

[knotty moth]

pitch can vary between different kind of voices, just tune the pitch shift as needed

[sick wraith]

i cannot stop getting repeating audio

[hearty peak]

-rt

[azure marsh]

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

[unkempt basin]

that worked

[shadow sparrow]

anyone else getting the "504 gateway time-out" thingy for applio colab?

[low shard]

gradio is down

just checkmark this

[shadow sparrow]

👍

[rare goblet]

Ayo? @shadow sparrow level 3 !!!

[low shard]

#📰│dev-updates message

[raw cloud]

whenever I talk in the google colab client of rvc nothing comes out

[odd shale]

None, there's no limit for dataset minutes.

[wary peak]

HOW CAN I DOWNLOAD THIS VS CHANGER

[rare goblet]

Ayo? @wary peak level 1 !!!

[odd shale]

Voice changer?

-rt

[azure marsh]

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

[odd shale]

Read the Deiteris docs.

[wary peak]

THANKS BUD

[odd shale]

Nope.

Altho if you use big datasets that are over 40-50 mins or 1 hour longer you could try using 9 or 10 on batch.

But 8 tends to be the most stable one for everything afaik.

So i recommend using mostly 8.

Probably their datasets are very tiny or don't know how batch works.

You're welcome bud.

I don't think there's a way to change it.

I don't recommend using other pretrains aside from the original one.

Are you making sure to use clean audio?

Nah, it's okay. Just asking.

MVSEP and UVR Online pages.

Then i apply cleanup on RX10 too.

If my dataset is over 10 mins, i use 8 on batch size.

Kaggle Mainline.

There you have.

-kaggle

[azure marsh]

📘 Kaggle Notebooks

• Applio Notebook, by Vidal Kaggle
• Applio Notebook, by Shirou Kaggle
• Music Source Separation, by Shirou Kaggle
• UVR5 NO UI, by Eddy Kaggle
• RVC Mainline, by Hina Kaggle
• Original W-Okada's Voice Changer, Kaggle
• Modified W-Okada's Voice Changer, Kaggle
• 🆕 UVR5 UI, by Eddy & ArisDev Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline Kaggle by Cauthess

Note: Kaggle limits GPU usage to 30 hours per week.

[odd shale]

Read the Cauthess one.

[azure marsh]

💻 (Local) Realtime Voice Changer

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[molten ginkgo]

-colab

[azure marsh]

☁️ Google Colabs

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

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

[errant flame]

just installed and wanted to try applio but it says no module named 'gardio' when executing run-applio what do i do ;-;

[tight tiger]

when i use rvc it just keeps cutting out how do i fix it?

[tight tiger]

-rt

[azure marsh]

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

[low shard]

#📰│dev-updates message

[feral marsh]

i still need help with this..... im using my gpu as usual. is there any other way i can determine this? the delay used to be like a second, now its 3 seconds