✨│ai-help

[spiral whale]

I've removed the instrumentals and am now trying to remove backing vocals from the tracks

Edit: got some help on general channel. Needed to choose process method VRA instead

[spice thorn]

no

XD

[proven hill]

okay(?)

[ember shore]

Hey guys, I want to make a voice model now, but where is the link to use titan?

[analog current]

What is the best colab notebook for training a model?

[rare goblet]

Ayo? @analog current level 1 !!!

[brittle wing]

what is the minimum requirments to train a rcv

[rare goblet]

Ayo? @brittle wing level 2 !!!

[rancid flint]

instead of having an option to select my GPU, I only have GPU(dml)cpu-gpu0-gpu1 etc. Im using the directML version because of amd, help would be really appreciated

[violet heron]

GTX 16 series or later, but you can train using colab

[tidal trail]

applio or mainline colab

[violet heron]

You would select which one is your AMD GPU

So like if your Gpu in task manager was gpu 1 you select that

[uneven nacelle]

would upgrading from an rtx 2080 to 4080 make much of a quality difference in terms of the voice or just in the speed at which its being processed?

[rancid flint]

Its GPU 0 apparently, thats the only one I have

[violet heron]

Then select GPU 0

Wait

[rancid flint]

thanks to another helper, I have been told to use wokada

[proven hill]

2080 is more than enough a 4080 would only improve latency

[violet heron]

If you have a AMD GPU it’s recommended to get the modded version of it

[rancid flint]

it is amd yes

modded version as in, the directML ?

[violet heron]

No there’s a fork that improves performance on AMD https://github.com/deiteris/voice-changer/releases/tag/b2120

“voice-changer-windows-amd64-dml.zip”

[rancid flint]

which one is it?

[violet heron]

“voice-changer-windows-amd64-dml.zip”

[brittle wing]

You mean i can train the rcv using Applio colab and get some good quality results

[violet heron]

Yes, but local is recommended over colab if you have the right hardware

[rancid flint]

Im downloading it right now, thank you a lot. What is the directMl used for then if you dont mind me asking?

[violet heron]

AMD but the one I linked is a fork that increases performance

Since the main version is iffy for AMD

This one should make it run better

[rancid flint]

is it supposed to open in a browser?

[wispy lodge]

Yes

[rancid flint]

also yes, it finally did let me select my gpu

[brittle wing]

i have a gtx 760, so what is the best way to get a really good voice model since i can t train it localy?i have the dataset

i mean ai voice

[violet heron]

That wouldn’t cut it, you will have to use the colab

[rancid flint]

It doesnt let me use the voice, I press start and then try to record it but nothing happens

[rare goblet]

Ayo? @rancid flint level 2 !!!

[wispy lodge]

You should select a model first

[rancid flint]

I did

[violet heron]

You need to select a model

And you need to click start

Not record

[brittle wing]

it is someone here that can make it for me, like payd job?

[wispy lodge]

Well, the model is not selected rn on your ss

You need to click it

[violet heron]

And to make sure it works set monitor to your headphones/speakers to see if it works

[violet heron]

#1191429836321849435 #1159289738314919936 you might want to consider training it yourself tho

[brittle wing]

i will try that if you tell me how long the vocals to be in dataset

[rancid flint]

I used another model and it does let me press on it

Ill test it now

[violet heron]

15 minutes is usually the sweet spot

[brittle wing]

like 5 vocals of 3 min?

[violet heron]

I’m saying 15 minutes worth of audio

It can be split up but doesn’t have to

[brittle wing]

the material i have it s segmented

[violet heron]

How long is it total?

[brittle wing]

13 min

[rancid flint]

is there a way to find out exactly what chunk is the best I could use?

[wispy lodge]

As long as perf number near the portrait is lower than the chunk you choose - it's a good value. perf is the time it takes to process the audio for your GPU

[brittle wing]

i can edit the material to make it one audio and make it in total of 15 min but..i think editing that will cut the quality

[violet heron]

You don’t need to

Having it split is fine

[brittle wing]

thanks mate

[rancid flint]

I found a chunk that was 30ms above my perf, but it sounded pretty bad, words were not clear at all although there was no delay, Is it a chunk problem?

[wispy lodge]

You mean it lags/cuts off or what?

[rancid flint]

yes, both

[wispy lodge]

Then stop voice conversion and start again. This can happen if you change chunk when conversion is running

Also in DirectML version, it's required to switch device to CPU then back to GPU if you change chunk or extra

[rancid flint]

what is extra for?

[wispy lodge]

It's an audio history. It improves converted voice quality at the cost of a bit more GPU usage

Extra at 2.7s is fine in almost all cases

[next lake]

can somebody help me get a villiger voice

[rancid flint]

never mind

Its an issue with my input

[wispy lodge]

You should use the same audio API for both input and output, like MME or WASAPI. With WASAPI, you should also match the devices sample rate, it's mentioned here in more details
https://rentry.co/Realtime-less-delay

[marble forge]

how are songs like this made?:
https://www.youtube.com/watch?v=QIoJdw40dQU
https://www.youtube.com/watch?v=-cvcwIWYjzs

[rancid flint]

thank you for letting me know

what are these about exactly? which one should be used?

[proven hill]

those are pitch extractions methods, use rmvpe_onnx or fcpe_onnx

[wispy lodge]

rmvpe_onnx usually works best in terms of performance/quality
fcpe_onnx provides better performance and less vram usage with around the same quality as rmvpe_onnx
crepe_full_onnx may provide best quality but can require much more GPU usage compared to other methods
crepe_tiny_onnx is the smallest and fastest one and provides overall good quality, but is very inaccurate in some cases

rmvpe_onnx and fcpe_onnx are usually most optimal

[rancid flint]

thank you guys

[violet heron]

Instead of dio use RMVPE

input should be your microphone

[pastel oak]

Honestly with a 750 ti not sure if its worth trying

[violet heron]

Output should be your audio cable if your planning to use in other apps

[violet heron]

[next lake]

but now how do i add custom voices

[rare goblet]

Ayo? @next lake level 2 !!!

[violet heron]

Edit button

[pastel oak]

Basic logic to click upload on an empty space xd

[violet heron]

RVC

[next lake]

then?

[violet heron]

Select the models pth as the “model”

And the models index for the index

!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.

[violet heron]

Read the line about “being productive”

[next lake]

ok

[rancid flint]

what could be the reasons why the voicechanger works well generally but not in discord?

[rare goblet]

Ayo? @rancid flint level 3 !!!

[wispy lodge]

VB Audio Cable may be the reason. It's usually better to use VAC

[wispy lodge]

750 Ti won't with CUDA version iirc. Feel free to try though

[rancid flint]

could you please give me more details about vac?

[pastel oak]

Downlaod link can be found here

https://rentry.co/VoiceChangerGuide#virtual-audio-cables-mandatory

Generally its the same as VB Cable except it works

[next lake]

can we hop on vc or something beacuse im dumb @violet heron

[violet heron]

https://rentry.co/VoiceChangerGuide there’s a guide for a reason

[rancid flint]

can you please guide me a bit here

[violet heron]

Setup64

[rancid flint]

I extracted everything and got this

[rancid flint]

setup64 again?

[violet heron]

Run setup64

[rancid flint]

I suppose its done

do I just change the output everywhere?

[violet heron]

Open the audio properties and make it not the default playback and recording devices

You don’t want that

[rancid flint]

it still sounds pretty bad on discord, any reason why maybe?

[violet heron]

The audio quality on discord sucks

[wispy lodge]

Like how? Could you show your current settings?

[rancid flint]

nothing is clear, barely anything can be heard and sometimes its like everything I say lags

[wispy lodge]

Should be working fine 🤔 Could you try switching to client audio?

[rancid flint]

also I keep getting this sometimes

[rancid flint]

which one do you mean?

[wispy lodge]

You should check command line that opens when you start the app

It may show the error

[wispy lodge]

I mean, you currently use server audio. Change it to client

[marble forge]

how are songs like this made?:
https://www.youtube.com/watch?v=QIoJdw40dQU
https://www.youtube.com/watch?v=-cvcwIWYjzs

[rancid flint]

Its the same, I have no idea whats wrong, could it be the mic?

[wispy lodge]

Not sure... Did you try restarting voice changer?

[rancid flint]

yes multiple times, because of the error too

[wispy lodge]

Hmm, could you make a recording of what it sounds like?

[raven wing]

waht do i check here

[wispy lodge]

And do you use index?

[rancid flint]

im not no

also im trying to record now

[wispy lodge]

But do I understand correctly that if you set output to speakers, it sounds fine?

[rancid flint]

actually when I tested it in a call, it sounded much better than the test I did before alone

but the quality is still kind of bad

[rare goblet]

Ayo? @rancid flint level 4 !!!

[next lake]

if it is some delay on the vcclient is there a way to fix it and how?

[wispy lodge]

Like still cuts out?

[wispy lodge]

And did you try checking this way?

[rancid flint]

yes, when my output is set to speakers it sounds much better

than before

[rancid flint]

yes exactly

that might be a mic problem?

also there is an echo

sometimes

[wispy lodge]

Shouldn't be. Usually this is caused either due to performance problems (high cpu usage, or gpu is slower than chunk size, check task manager) or due to vb audio cable

[wispy lodge]

Echo occurs when your microphone picks up the sound from your speakers

[rancid flint]

my gpu and cpu seem to be fine

my mic is not the best though, thats why im considering that

I tested it on multiple mic test sites and its actually really good compared to discord?

[wispy lodge]

Huh, that's weird. Not sure what could be the problem here...

[next lake]

how to i connect vcclient to roblox

or what is the device namn called so i know

[rancid flint]

yeah It was fine for a while now its back to cutting out

Im getting a new mic anyways and Ill test and see

Im almost sure thats the problem

[frank palm]

One message removed from a suspended account.

[odd shale]

Which model are you using? (Nvm i've looked at the models channel and ur using Blyuv's softgirl model)

Umm.. Make sure you're using settings according to your GPU.

Take a look at the W-Okada guide.

https://rentry.co/VoiceChangerGuide

[frank palm]

One message removed from a suspended account.

[faint wraith]

wow fondateur de purdir?

🫢

wowowo

i am fan

[odd shale]

What are you doing?

[faint wraith]

nada

[frank palm]

One message removed from a suspended account.

[rare goblet]

Ayo? @frank palm level 1 !!!

[faint wraith]

soy spanish

[odd shale]

Qué haces?

[faint wraith]

è italiano

[frank palm]

One message removed from a suspended account.

[faint wraith]

pycc

ok sorry stop i just wanted to say malik is cool

[frank palm]

One message removed from a suspended account.

[faint wraith]

np PurDir is cool

[frank palm]

One message removed from a suspended account.

[faint wraith]

i heard of it a lot

[frank palm]

One message removed from a suspended account.

[pure needle]

I need help, how do I do covers? I searched for the voice I wanted, and it asked me to download a .zip of the voice, where do I put this .zip? how do I do?

[low shard]

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

if you got a good pc do it locally, else do it online on cloud

[slim wadi]

just curious where i could go to make my own model? also wondering where to start or if i only need audio of voices i want to use?

[lunar ice]

My god I feel stupid, I don't even know how to do this step by step hahaha

[violet heron]

-docs

[azure marsh]

Suggestions for @slim wadi

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[pliant wigeon]

hi guys, i havent used rvc for a while, and now it just freezes

cant even run it

any suggestions

[knotty moth]

elaborate the problem? other than possibly the disk failure

[pliant wigeon]

it just doesnt respond at all, it used to automatically just go

may i dm u

it just freezes and doesnt go, on the cmd it just says found gpu nvidia geforce rtx 3090

[knotty moth]

no, here, let others help you maybe

[pliant wigeon]

it doesnt let me send a pic but its basically the go-realtime-gui not responding

[rare goblet]

Ayo? @pliant wigeon level 1 !!!

[knotty moth]

without any error messages and something else, I have no clue
anyway for RVC realtime, you can go here https://rentry.co/RVCRealtimeGuide

[pliant wigeon]

ah ty, i didnt see they had an okada update

im using one from like years ago lmao

tysm!!!

[rare goblet]

Ayo? @main tusk level 1 !!!

[misty elk]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[paper dock]

titan pretrain sounds alot better

v10 has the same/more cleaner dataset than v9

both trained on 40k

v10 at 540 epoch and v9 at 500 epoch

v10 had batch size 12 while v9 had batch size 7

thoughts?

[brittle wing]

alguien que hable español y me ayude a configurar se sigue escuchando mi misma voz

[rare goblet]

Ayo? @brittle wing level 2 !!!

[brittle wing]

[analog obsidian]

tienes el virtual audio cable instalado?

[brittle wing]

al parecer creo que no

[analog obsidian]

-realtime

[azure marsh]

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

[analog obsidian]

dale click al primer link

ahi tienes todo el tutorial

si no entiendes solo usa deepl para traducirlo

[brittle wing]

ok gracias

se sigue escuchando mi voz y eso que tengo todo instalado que debo hacer 😦 .?

[analog obsidian]

input > tu microfono real
output > line 1
monitor > nada

dejame abro wokada y te muestro

[brittle wing]

vale

[analog obsidian]

primero, cambia el f0 de crepe a rmvpe

[brittle wing]

ya

[analog obsidian]

esto se tiene que ver asi (en input pones tu microfono)

en chunk pone 112

y extra 16384

no actives la opcion que dice passthru (que no este en rojo)

[brittle wing]

ya conecte mis audiculares que son cableados y ya puse los demas ajustes

no esta activado

[analog obsidian]

reiniciaste el pc luego de instalar vac lite?

o bueno, laptop

[brittle wing]

no

[analog obsidian]

reinicia

[brittle wing]

ok

[analog obsidian]

luego de reiniciar tienes que ajustar el tune
esto depende que tan diferente es tu voz del modelo, por lo general, si eres hombre y quieres usar un modelo con voz de mujer, usa tune 12

[brittle wing]

ya esta

espera ajusto

[analog obsidian]

para grabarte puedes usar esto
en output pones tus auriculares (no el mic)
y luego le das click a start, cuando quieras parar la grabacion le das a stop, y podras oirte
otra opcion tambien es grabarte en audacity, ambas funcionan

y cuando quieras usar el voice changer en discord fijate de usar line 1 como tu microfono principal

[brittle wing]

ok gracias voy a probar

en esto no tiene nada que ver

[analog obsidian]

en start pone 0.2, en end pone 0.8

trancate 300

silencefront off y luego le das close

[brittle wing]

mas tarde lo intentare de nuevo tengo que dormir

[delicate wind]

how to fix this

NameError Traceback (most recent call last)
<ipython-input-2-a8064234ea5f> in <cell line: 20>()
18
19
---> 20 if mode == "elf":
21 # !LD_LIBRARY_PATH=/usr/lib64-nvidia:/usr/lib/x86_64-linux-gnu ./vcclient_latest_for_colab cui --port {PORT} --no_cui true &
22

NameError: name 'mode' is not defined

[violet heron]

Which file are you opening and on what platform

[dull orchid]

Is there anything that you can install local that is the same as Suno ?

[night karma]

why does the voice sound static

[tame mica]

how do i uhhh train 32k in kaggle

[glad zealot]

Press v1 then v2 again

[tame mica]

nice thanks

[cold sentinel]

why my shi sound so bad

cutting out n shi on discorrd nd shii

[hybrid wagon]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[rare goblet]

Ayo? @hybrid wagon level 1 !!!

[oblique cliff]

how to i prepare applio to train a voice?

[odd shale]

Prepare Applio?

Actually if you would like to train a model, you need to make a dataset of the person/character you want a model of

Read the docs there:

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[odd shale]

!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.

[violet heron]

Just in other apps like discord or also in the voice changer itself?

[cold sentinel]

only apps

mostly just discord

[violet heron]

Probably your audio cable, use VAC LITE instead of VB Audio Cable

https://software.muzychenko.net/freeware/vac470lite.zip run setup64

Make sure to make it NOT your default mic/speakers

[boreal sage]

i already ran feature extraction

[brittle wing]

can anyone guide me here

[brittle wing]

Hello, I wanted to create my own voice model and use it for music. Could someone help me or write down the steps on how to do it?

[brittle wing]

i iallready installed tensorBoard but at return i got Unable to convert function return value to a Python type! The signature was () -> handle

[rare goblet]

Ayo? @brittle wing level 3 !!!

[terse siren]

this ^^ i need help doin this too

except for casual talking

[lunar ice]

Do you have a tutorial on how to mix the cover?

[knotty moth]

there are many FL studio tutorials for mixing you can search for

[lunar ice]

thanks

[jaunty shale]

??

why is this happening

anyone have any idea?

[hard aspen]

-realtime

[azure marsh]

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

[plain saffron]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[rare goblet]

Ayo? @plain saffron level 3 !!!

[gilded crescent]

Hello .
i was wondering if anyone can help me with my Voice changer problem . no matter what voice model i apply . i choose Start and then the option in right which activate the Voice changer . but in the end i still hear my own Voice as outcome . no voice changing . the in and out option which is to raise or lower the voice works just fine but it doesnt change the voice at all and on all models i still hear my own voice . it used to work before but now it dont.

[violet heron]

What’s your input/output/monitor

[gilded crescent]

input my mic . output is the cable . and monitor is the headphone

i did everything people in the youtube said so the settings are right . i followed everything they mentioned in the video step by step thats why im confused why it play out just my own voice

[rare goblet]

Ayo? @gilded crescent level 1 !!!

[violet heron]

What’s your GPU

[gilded crescent]

Nvidia 1030

it was working before

[violet heron]

That won’t really work

[gilded crescent]

Thanks

[rain anchor]

pls help RVC working on cpu but i choose my gpu, how to fix this?

[violet heron]

What’s your GPU?

[rain anchor]

3060 ti

[rare goblet]

Ayo? @rain anchor level 1 !!!

[violet heron]

Are you using w-okadas or RVC

[rain anchor]

rvc

[hard aspen]

Hello, could someone help me? When I place a voice and configure it, it takes a long time to speak (I have it at 112) and it is short, the same thing doesn't happen to me with the default ones, what could it be?

[violet heron]

That is w-okadas

[rain anchor]

https://media.discordapp.net/attachments/1010351049179148388/1070922579046563910/1670312087349.gif

[gilded crescent]

whats the difference between W-okadas and RVC ? because mine is same as His

[violet heron]

RVC is audio to audio and runs in a gradio ui

W-Okada does use RVC models

[gilded crescent]

oh . now that makes sence . thanks again

[storm holly]

how well is the autumn rvc gonna work if i have a 24 second database(and 2/3rd of that isn't even autumn but autumn's VA sounding like autumn)

[feral ivy]

how do i make a mp3 of a famous artist singing one of their songs but with the lyrics changed?

as a complete beginner

[spare shale]

https://www.weights.gg/ + https://www.suno.ai/

[jaunty shale]

can anyone help me with this? trying to do model for 3 hours

[brittle wing]

@crude bolt i need help, can u friend me

[thin stump]

That's a sign to stop using rvc disconnected

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[thin stump]

The mainline colab guide has a kaggle link included, which I highly recommend

[crude flame]

Did the thing. These samples were made with the voice changer.

[glacial pollen]

Neatttt, now, what do you think of it?

tested the generalization ability?

[crude flame]

i havent formed an opinion yet

[analog obsidian]

they sound almost exactly the same, but i noticed the codename finetuned has better vocal fry

[glacial pollen]

Quite possibly features are more baked in

Is slightly more stable too, to my ears

[analog obsidian]

yeah i also noticed this too

[glacial pollen]

but biggest deal would be testing out it's generalization and pitch handling

[analog obsidian]

razer did u trained in fp32 as well?

[glacial pollen]

try to push the model to the limits

use some male vocals input

crank up the range by 2 octaves

so, 24 for transpose

maybe even 36

or try all: 12, 24, 36

for each trained model

[crude flame]

im like 78.58% sure i did

[glacial pollen]

finetuned handles it a lil better

So yeah, 2-step training for the win so far

stage 1: transfer-learning and preparing base

stage 2: actual finetuning to semi-adapted gen and disc + seen speaker

which adds in possibility to re-learn badly learned representations or focus on non-learned parts

✨

[analog obsidian]

this method only can work in bigger datasets and not smaller datasets? like if i try training g and d of a 10 minute dataset, and then use those g and d to train the same 10 minute dataset again, its not going to make a significant change vs training with the original pretrain?

[glacial pollen]

Hard to say to be honest as it's, like in most ml, matter of case but

logically, you're most likely right on the small set part

less risk of an actual overfitting and biasing

but more tests need to be done really, on both og and custom pretrains, featuring various sets

say: 1, 3, 5, 8, 10, 12, 15, 20, 25 m sets

etc

[analog obsidian]

im gonna try what i said to see if does make a change vs original pretrain

[glacial pollen]

Alrrr, report any findings if you may

good luck ✨

[analog obsidian]

thanks!

[sullen bobcat]

I understand the Ai Changer wont be perfect but is there a way to improve the quality.It sounds perfect if i keep my sentances short like 3 words or so but anything longer it will go robotic? Is there a way to improve that

[analog obsidian]

might be a model issue (probably badly trained), or it could be extra chunk size too low

[sullen bobcat]

okay thanks ill increase chunk size and see how that sounds if not guess im paying some money for a model

[analog obsidian]

you have to increase EXTRA btw, not regular chunk size

extra handles the model's clarity

[sullen bobcat]

okay yeah cheers

[analog obsidian]

so if its too low its gonna sound too low quality/robotic

[sullen bobcat]

turns out i had it a bit too high.I had my extra at 16k but 8k sounds perfect for me its less robotic

[analog obsidian]

oh, as long is working fine, you can keep that value

[viral flame]

can someone help me with my ai song cover

every time i try to make the cover my pc crashes or it says it failed

nvm i guess

[rare goblet]

Ayo? @viral flame level 1 !!!

[violet heron]

What’s your GPU

[viral flame]

uhhh

not too sure about that

does rvc support japanese vocals?

[violet heron]

Yes

[violet heron]

Check in task manager under performance and GPU 1

[viral flame]

alright

radeon RX 570 series

thats what it says

[violet heron]

AMD GPU which so conversion would take a bit

[viral flame]

also

rvc puts it on cpu and i cant switch it to gpu if that has anything to do with it

for processing mode

[violet heron]

You can’t use a AMD GPU on RVC

Since RVC uses Cuda

[viral flame]

ah

[violet heron]

If you want faster conversion I would recommend using Ilaria RVC which does conversion with a powerful ZeroGPU in the cloud

-spaces

[azure marsh]

<:huggingface:1179800228946268270> Hugginface Spaces

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

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

[viral flame]

ok thanks

[rare goblet]

Ayo? @viral flame level 2 !!!

[red kayak]

This?

[storm holly]

rvc disconnected fucking up again

[glacial pollen]

oh, nono

pretrains are needed

[storm holly]

says i need a GPU

[glacial pollen]

the og needs or whatever as a base for transfer learning

[red kayak]

Ohh

[glacial pollen]

the 2nd run ( on the model's G/Ds ) is the targetted fine-tuning phase

[analog obsidian]

xD

[red kayak]

Train first with og

Then take those from ur model

[glacial pollen]

yes

that

[red kayak]

And train with those

Oke

I'll give it a spin

[analog obsidian]

wait

so is 2 times

training two times

i thought it was 3 times xD

[thin stump]

Just use the big three; Applio colab, ilariarvc, or mainline

[storm holly]

applio doesn't work for training

[thin stump]

https://rentry.co/RVC-Mainline-Kaggle if you click on the mainline colab hyperlink, that'll take you to the colab guide. They're the same thing

[storm holly]

i think it just feels i'm using the thing too much

[thin stump]

mhm, rvc disconnected constantly has bugs whenever someone changed the code slightly or pushed a new feature

the mangio fork is outdated anyways

not necessarily new, but I member' when the enable autosave feature broke the tensorboard

[uncut onyx]

Might anyone know why RVC won't work with my nvidia rtx 2080? I select it as the gpu, and nothing happens when I click start. But if I use CPU, it works, but sounds bad

[knotty moth]

which RVC?

[uncut onyx]

vcclient_win_cuda_2.0.40-alpha

im updating my driver now to see if that helps

[knotty moth]

that seems broken, try this https://github.com/deiteris/voice-changer

[uncut onyx]

Thank you! I'm going to try that one.

[rare goblet]

Ayo? @uncut onyx level 1 !!!

[uncut onyx]

thank you bot

[proper pivot]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[steel kite]

The audio isn't uploading to the webUI when I try to upload the file, how do I fix it?

@hearty idol @crude bolt @tired kraken @long forge @analog obsidian @scenic gale @steel forge I can't ping the helpers role for some reason

[acoustic scarab]

what webui to be precise?

[steel kite]

Hina Mod

[acoustic scarab]

like, google colab?

[steel kite]

yes

[acoustic scarab]

am not a colab user, sorry

[steel kite]

So who's supposed to help me before the collab expires?

Also how do I ping all the helpers instead of just doing it one at a time?

[barren canopy]

how about you...

don't?

[steel kite]

Restarting the collab

[rare goblet]

Ayo? @steel kite level 3 !!!

[steel kite]

Why not? I'm just asking for help?

[barren canopy]

Then just ask and wait for an answer

Tagging all helpers is just not the play

Ever

[steel kite]

Sorry, I just needed help before the collab GPU timed out

[steel kite]

Still not working

Idk how to fix it

Trying it on Chrome now

Still not working

Trying with a VPN

Still not working

[thin stump]

Have you heard of huggingspace before? Colab isn't the only platform https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit

[steel kite]

Yeah apparently it's not connecting to a runtime despite, the fact that I haven't used collab in weeks if not months ago

[steel kite]

So how do I use it through the cloud and not locally

[thin stump]

click the links. You can't get lost

[steel kite]

Waiting for the vocals to upload. Wish it had a loading bar

[steel kite]

Still loading, idk what's wrong with it

[thin stump]

There is nothing wrong. It's a you problem since it seems like you're rushing to get this done

[steel kite]

Not really, I'm just kinda worried that it's not actually working

Idk if there's a time limit to this one like Collab is

[thin stump]

even so, huggingspace has more gpu than colab without the stupid time limit

[steel kite]

Oh good

Okay yeah the audio uploaded

[small plank]

my voice changer client js has a white screen lol

[severe thicket]

How to produce momoi voice

[steel kite]

How do I adjust Crepe hope length on Ilaria RVC Zero?

Also does the quota reset after a day like Collab does?

[proper pivot]

-colabs

[azure marsh]

☁️ Google Colabs

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

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

[jaunty shale]

why?

[thin stump]

Well if its been fixed for you then disregard that message

[thin stump]

either way, I can't force you to read the guide I linked earlier because it's the very first thing you would see (regarding the mangio fork). I hear you loud and clear ❌

[jaunty shale]

i know how to use it, its just that this never happened to me before

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[rare goblet]

Ayo? @ornate marsh level 1 !!!

[thin stump]

Yeah it's just the dependencies. What I'm referring to is this section here on the mainline rentry guide explaining why rvc disconnected isn't ideal for training anymore

If you don't care about the constant bugs or syncing graphs, then keep at it

[languid lotus]

-collab

-colab

[azure marsh]

☁️ Google Colabs

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

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

[low shard]

The ZeroGPU quota resets around after 2 hour, but that depends, you should get a 'retry in ...' when u finish your quota

[low shard]

Like:

• make ai cover
• make voice model
• use voice in realtime for calls
  ?

[sand quarry]

heyy so i get an error but i have no idea whats wrong

is it bc of a data set or what

[sand quarry]

what do i do

[brittle wing]

HY ASSOCIATES ,FRIENDS AND ENEMYES , strong question, when you do colab after install and run that shells to run applio you go to public url and that it you can do you work ?

[night tree]

can someone tell me how much epochs should I train for 7:15 minutes of dataset? and which what method (harvest, rmvpe, pm, rmvpe_gpu) ?

please

[sand quarry]

rmvpe and i dont know abt epochs its just how much model has to be trained u can do what ever but you can over trained it so i always do 100,150,or 200

thats how i do it

[jaunty shale]

i tested it again, it works now

[visual quarry]

Hello guys, I'm using a pretrained model which I found in the #🔍│find-models, but after doing the "download model" thing, the new model is simply not appearing after hitting "Refresh". Any help would be great

[violet heron]

RMVPE

[night tree]

thank you bro

[rare goblet]

Ayo? @night tree level 4 !!!

[plain fulcrum]

Is colab not working for anyone rn?

I swear this the most fragile site ever😭

It's tellin me they cant gemme a gpu💀

[storm holly]

i knew disconnected would come back eventually

[storm holly]

how long should i train a 30 second database

[violet heron]

That won’t be enough time

[rare goblet]

Ayo? @violet heron level 104 !!!

[violet heron]

[rare goblet]

AI HUB's leaderboard

Want to view more than the top 10 users? Click here

[storm holly]

trust me, it's better than the original database which was 10 seconds

[pastel oak]

check tensor until you OT

[violet heron]

You still won’t get a decent mode

[brittle wing]

how log should i train a 17 min database

[pastel oak]

check tensor until you OT

[soft stratus]

I had a problem

Do you know a new update for easygui

[pastel oak]

-colab

[azure marsh]

Suggestions for @soft stratus

☁️ Google Colabs

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

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

[soft stratus]

But i had a problem

And my dataset said you don't ahve

[pastel oak]

I dont know, use something else

easygui is probably outdated

if u try to train a model check out rvc disconnected or ilaria

[soft stratus]

Are you sure

[pastel oak]

easygui hasnt been used in a while

[soft stratus]

I always use it

[soft stratus]

My problem is this

[pastel oak]

ok well dataset is empty it says so link is wrong or folder has nothing

Try different link or folder

[soft stratus]

I tried but no way @pastel oak

[dull orchid]

Hey, is there anything you can install local, that is kind of the same as Suno AI ?

And is there somekind of AI pitch correction tool ?

Something like Melodyne but than AI

[elfin salmon]

how delete Rvc models in apillo?

[violet heron]

Delete the .pth and .index files

Should be in weights and/or logs folder

Might be slightly different because applio

[vestal fox]

lul

there it show nothing

cant send a pic here

[violet heron]

Once you get level 1 you can

[vestal fox]

oh okaz damn

can i get somewere help-

?

[sand quarry]

hey so i still have this is it a bug or something

plz help

[vestal fox]

pls help me

[rare goblet]

Ayo? @vestal fox level 1 !!!

[vestal fox]

ahahha

its white for me

and this

[pastel oak]

1. What GPU do you have
2. Do you have antivirus program

[vestal fox]

is it right?

[pastel oak]

Yes

Installing files

[vestal fox]

ah okay than sry

ty

[obtuse stream]

does anybody has bohemia rvc model?

[hollow dune]

Can somebody just convert my audio for me😭

[sand quarry]

what do you mean

[hollow dune]

My PC broke and I wanted to make a JW song 😭 so I need someone to just run it for me then send the file

[sand quarry]

nah but whta do you mean run it

like in rvc

[hollow dune]

Yh bro 😂

[sand quarry]

im too dumb what do you want to convert

[hollow dune]

An audio file/recording

[sand quarry]

with?

[hollow dune]

RVC JW model

[sand quarry]

who you mean jw cuz again im dumb

[hollow dune]

Juice WRLD

[sand quarry]

okay priv chat

[hollow dune]

Ur not accepting requests rn 😭

[rare goblet]

Ayo? @hollow dune level 2 !!!

[sand quarry]

Ur not accepting request too

Wait I'll try in the setings

[hollow dune]

Oh my bad 😭

[hollow dune]

Try now

[sand quarry]

Done

[rare goblet]

Ayo? @sand quarry level 4 !!!

[sand quarry]

U have the model ?

[low shard]

You can search rvc ai voice model:

• #1175430844685484042
• #🔍│find-models
• https://weights.gg/
• 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/

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.aihub.wtf/

[limpid garnet]

hi! this is my first time using rvc, yet i cant make it work properly, any help?

i installed localy, my audios come like if they were zombies lol

[pastel oak]

did you adjust the pitch/tune

[high grove]

travis scott

[sand quarry]

please it happens everytime

[thin stump]

-colab

[azure marsh]

Suggestions for @sand quarry

☁️ Google Colabs

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

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

[thin stump]

You're in for a bad time if you use anything with the Mangio fork

[violet heron]

So true

[inner moat]

best rvc?

i dont rememver

which one

that is on youir pc btw too

[dusty garden]

can someone help me

when i talk my friend says he hears my real voice then the voice mod

how do i fix that

[hot lagoon]

When I tried to convert synthesize vocals and talk about synthesizer v into RVC vocals are too quiet even though I’m pretty sure I changed the gain a couple of times, but it’s still too quiet

[cursive knot]

I unfortunately don't know how to fix that, but I do know of another free ai voice changer on github that allows you to use your own AI models as well as any of the ones available from this server or on the web

[rare goblet]

Ayo? @cursive knot level 1 !!!

[cursive knot]

wow, so exciting!

[cursive knot]

nvm, it seems like someone else already helped you with that exact one

[misty elk]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[hot lagoon]

can someone pls help

[storm plinth]

Is there any guide how to use colab?

[rare goblet]

Ayo? @storm plinth level 1 !!!

[slim wadi]

is there a good range for minutes of voice to things like epochs? like is 15 minutes of voice 200 epochs or is there any conversion to use?

[hidden basin]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[pastel oak]

No you check tensorboard until you OT

[slim wadi]

ot?

[pastel oak]

Depending on what you need

https://docs.aihub.wtf/

[pastel oak]

overtrain

[storm plinth]

I just need how to import model

I made vocal

[slim wadi]

into w-okada?

[pastel oak]

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

[storm plinth]

What means w-okada?

[pastel oak]

which colab

[storm plinth]

I think uvr5

[pastel oak]

Thats a vocal remover app

[slim wadi]

so either learn to understand tensor board and or guess and check?

[storm plinth]

Which colab should I use?

[pastel oak]

Its insanely easy once you read through it once

[slim wadi]

i just did i hate to say it but im just really really slow 😭🫶

[pastel oak]

#✨│ai-help message

the very fast inference if u want

[pastel oak]

Train, save model file every x epoch you chose
Look at g/total
wait for it to be on its lowest point
keep training until youre sure its the lowest point
use model of lowest point

[slim wadi]

so lemme send out a “hypothetical,” i have 20 minutes of data set and i have it on every 25 epochs save, look at the graph the way the link walks me through to until it dips, then how do i go to right before the dip so i can save it while it over trains. also do i just set it to say 1000 epochs and then watch the graph from there?

hope that makes sense lol

[pastel oak]

its automatically saved if you enable the save model every epoch thig

Dont remember the name

Set to 1000 ye

With 25 itll be less accurate so lowest point could be idk 15k steps but the one before it you have 13k and after is 18k youd still use 13k

[slim wadi]

should i set it to save every 15 or 10 then? i noticed when i first trained a model at 10 that’s what took the most time (the saving)

but then again how do i stop it and save at the epoch i want 😭

[pastel oak]

ye

[slim wadi]

using web version btw

in case it changes anything

[pastel oak]

nothing at all

I mean idk how many times to say it, you save models by enabling the button saying save blabla model to weights folder

you keep asking the same question xd

[slim wadi]

OHHHHHHHHH

WAIT

MBBB

I GET IT NOW

SORRY

BUT THANK YOU SO MUCH

[pastel oak]

Ez

[sand quarry]

I used every one of them nothing worked

[thin stump]

IlariaRVC should have worked. Use this guide for mainline since it's complete https://rentry.co/RVC-Mainline-Colab

[low shard]

btw the error was of the dataset https://discord.com/channels/1159260121998827560/1261469256558579732

[thin stump]

Oh I see, I don't see the benefit of training on rvc disconnected because of little errors like that . Most people skim over the aihub guide anyways

@sand quarry

ye you were told to stick with WAV or Flac files. Ogg is fine too

[sand quarry]

is mp3 ok too

[low shard]

wav & flac are the best

[low shard]

mp3 isn't lossless audio file format, its better to use WAV or FLAC (the only difference between those 2 is that flac compresses the data without losing quality while wav is purely the original data)

[sand quarry]

worked with mp3

thx yall so much

<33

[thin stump]

works but not the best quality. I'll link this here since it's the whole entire server log summarized with bits from #🔥│model-maker-chat https://rentry.co/RVC-dataset-RX11

[rare goblet]

Ayo? @thin stump level 32 !!!

[flat coyote]

does it mean the training failed ?

like it would get stuck at :Train Epoch: 1 [0%]

[flat coyote]

nevermind it turns out I only need to use one audio

[pastel oak]

Ehh no you just have to wait for the first epoch to finish

depending on your gpu can take longer

[jolly parrot]

IS there somebody who could help me install RVC on mac?

When I'm trying to run I get an error that has to do with ffmpeg

[odd shale]

🐢

[pastel oak]

did you follow the readme on github

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

[pastel oak]

🥬

[jolly parrot]

Followed a different tutorial, but will try this one too 🙂

[odd shale]

Yeah, i don't suggest you following YT tutorials.

These tend to be outdated

[jolly parrot]

Mmm it more or less follows the same steps, during install of ffmpeg I get errors

[rare goblet]

Ayo? @jolly parrot level 3 !!!

[odd shale]

But what do you want to do on RVC?

Make a model or use models?

[jolly parrot]

Both

[proven hill]

you need training?

[jolly parrot]

Training in how to install? Yes | Training models? No I want to do that myself 🙂

[proven hill]

no i meant

if you needed training in the mac installation

[jolly parrot]

Yes I do

[proven hill]

because iirc mac doesnt support training

[jolly parrot]

Oh fuck it doesnt at all? Does it support combining models at least?

[proven hill]

you mean merging?

[jolly parrot]

Yeah

[proven hill]

should do i think yes

[jolly parrot]

Alright, that would also already be nice

Then I guess I can train using google collab or whatever

And then combine locally

Unless I could combine on google collab as well 🙂

or sorry merging

[proven hill]

i think you cant do that on colab

[jolly parrot]

Oh damn, then I need to get it running on mac

Or I could dust off my old windows laptop, but it has a 1070, so that's gonna be slow as hell

[wispy lodge]

Models merging doesn't even require gpu

Not sure if training is feasible with 1070 though

[jolly parrot]

Oh I didn't know that

[jolly parrot]

Then I will train on collab, merge on my windows lappy

Thanks for the advice everyone

[harsh trellis]

can someone help me with RVC

[knotty moth]

the only reason to merge is to do it on the same model, different epoch, as explained in this guide #✨│ai-help message

[harsh trellis]

Traceback (most recent call last):
File "E:\RVC-GUI-pkg\rvcgui.py", line 129, in vc_single
audio_opt = vc.pipeline(
File "E:\RVC-GUI-pkg\vc_infer_pipeline.py", line 326, in pipeline
sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
RuntimeError: CUDA error: the launch timed out and was terminated
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
im getting this error whenever i try to convert an audio in RVC can someone help me?

[proper shale]

id suggest using something else, RVC-GUI is outdated really

https://docs.aihub.wtf/

RVC > Local section, use either Applio or Mainline

[harsh trellis]

thank you

[rare goblet]

Ayo? @harsh trellis level 1 !!!

[proper shale]

You're welcome :)

[plain fulcrum]

guys what does this mean?

im connected to the colab gpu

[proper shale]

you didn't run Set Training Variables maybe

[plain fulcrum]

yeah tht might have been why

i started over it seems to be working now

[proper shale]

yeah that was probably it

[zealous shuttle]

the output audio always like this when i do inference on Applio colab, any best settings for inference in applio?

on a bottom audio

[analog obsidian]

decrease protect voiceless consonants to 0.33

also set volume envelope to 0

[opal vortex]

how do i run it ???????????

[proper shale]

run what exactly

okada? applio? rvc realtime?

[opal vortex]

idk i just paid 54 dollars to some guy and he told me to download it

[past owl]

guys who can help me?

[zealous shuttle]

is it just me or the TITAN pretrain with rmvpe sounds more robotic?

[royal sinew]

Hello. Does this apply to this?

[zealous shuttle]

the dataset is around 13 minutes

[thin stump]

Titan has some aliasing artifacts created and most of the low end ringing noises are mostly from rmvpe

but rmvpe is the best because mangio crepe wants no noise at all. It's already hard enough to clean a dataset

[past owl]

?

[thin stump]

Don't use numbers for your folders or model name. All the lines above this are waiting for you

to change it

[past owl]

thank you!

you really helped

[plain fulcrum]

How do I resume my training on colab? I had my save frequency set to 10 epochs but I got runtime disconnected, how can I continue from the last epoch I saved?

[thin stump]

You lost your model if you didn't export it. Because save frequency only determines the .pth file checkpoints but it's not logged in your drive

Other colabs will have automatic backups but not RVC disconnected

That's why I recommended kaggle for 12 hours of runtime

[plain fulcrum]

My model's gone

[thin stump]

[plain fulcrum]

So what's even the point of having the save frequency😭😭😭

Omg

[plain fulcrum]

Whats that

İs tht like another colab

[thin stump]

RVC disconnected doesn't have a log folder. Another reason why it's outdated. Oh I linked this before https://rentry.co/RVC-Mainline-Kaggle

[plain fulcrum]

İs that local?

Cus I don't have a gpu I can't do local stuff

[thin stump]

It's also owned by Google. No it's on the cloud

[plain fulcrum]

Ohh so basically just colab2.0?

I'm liking the sound of this one

[thin stump]

Colab is their lil money making machine, so not really

Ppl don't know kaggle exists except IT students etc

[plain fulcrum]

I'm not even kidding I was debating buying colab pro cus of thier unusable free plan

İs it as simple to use as colab? Cus I already have my dataset I'm ready to just jump onto this kaggle thing

[thin stump]

Yeah I didn't know they gave so much gpus . The guide is almost complete but it's the same thing as the mainline colab guide I wrote

[plain fulcrum]

Already I'm sold on it

[thin stump]

If you click on the hyperlink, it'll take you to that guide

[plain fulcrum]

Lemme go try it out

[thin stump]

You just have to figure out the path file part because I don't have a dataset on hand

The colab guide is complete if you need a reference.

[plain fulcrum]

So I just paste the path of the dataset file?

[thin stump]

I think you need to type it in

For colab, it's copy and paste

[plain fulcrum]

this it right here?

this looks alot like colab

[thin stump]

Nope, did you click on the hyperlink in the guide?

It'll take you to the page

[plain fulcrum]

Ohh I didn't see it alr lemme check

this should be it

[rare goblet]

Ayo? @plain fulcrum level 5 !!!

[plain fulcrum]

i clicked the top link in the guide

or maybe im buggin again

[thin stump]

Yup just follow what the guide say. You probably wouldn't be lost because I provided a lot of images

Because the way I see it, I have to ask questions myself to figure it out so I'm just reposting what I learned

[plain fulcrum]

yeah thats fine ill try to follow the guide

thx so much for the help didnt know rvc disconnected was outdated lol

[thin stump]

Yeah that took forever for me to realize. Benefits of lurking on the server

[fallow orbit]

https://youtube.com/shorts/IU0VH0A5clk?si=dhP_cxfwIN34y34Y

[plain fulcrum]

wait a second

we already got a problem

it says enable the internet on the sidebar

where lol

[thin stump]

You have to make an account

[plain fulcrum]

i did

logged in and everything

[thin stump]

Provide your phone number for verification purposes

[plain fulcrum]

oh

didnt do tht yet

[thin stump]

And you'll have a sidebar

With the Internet toggle 👍

[plain fulcrum]

do i have to run a cell first or something?

i still dont see it

i verified aswell

[thin stump]

Refresh the page

[plain fulcrum]

is it supposed to be right there on the left?

[thin stump]

Don't worry about making mistakes because that's a shitload of GPU for the average person. Yuo

Yup

[plain fulcrum]

i refreshed still aint nothing happened

i think it might be an eyesight issue

see look

i dont see no internet

[thin stump]

I'm on mobile so I can't help you at the moment gtg

Session options

[plain fulcrum]

ahh damn

[thin stump]

Under your session options

[plain fulcrum]

OHHHHH

I SEE IT

[thin stump]

I'll update the guide to be more clear later

[plain fulcrum]

thx time for step 3

watch me get stuck again lol

[zenith lodge]

why won't my mic input audio anymore?

[plain fulcrum]

yup my brain cant handle this

officially

[dawn tendon]

Do i put voices here to put them on Inference later?

I forgot

[thin stump]

What are you stuck on?

Because you're just inputting the ngrok token and custom pretrain before you run the entire code

[dull wave]

what soft for real time voice change?

[fringe cave]

someone know how to fix this?

When i try to export to onxx: (AMD)
TypeError: Cannot read properties of null (reading 'path')

[serene wigeon]

what's the bot command that will bring up links to the goolge collabs that can be used for voice fitting with models

it would be nice to have it as a permenant link in some channel

-colab

[azure marsh]

☁️ Google Colabs

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

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

[serene wigeon]

this one

had to just scroll up a ton until someone happened to use it

[rare goblet]

Ayo? @serene wigeon level 7 !!!

[severe trellis]

-help

[azure marsh]

Command has expired

[severe trellis]

-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

Credits to Faze Masta and Antasma for compiling these links.

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

[gusty scroll]

Did you make this guide? If so, thank you so much!
After much research I reached the conclusion that using kaggle would be the best option for me. But due to my lack of time and knowlegde I've yet to try setting it up. So that guide should be of help 👍

[thin stump]

Yw! I'll add more to when I eventually have a dataset ready which should probably be next week. I was just setting it up for myself and will have to figure out the file path directory part soon

[zealous shuttle]

and..... What is the best pretrain for around 10-15 minutes dataset?

[thin stump]

Original or Titan . We have to wait until the RVC architecture changes so there's BigVGAN pretrains and not Hifigan

all the current pretrains just uses Hifigan as their base, I think

[zealous shuttle]

Ouhh.... I see

[thin stump]

#📰│dev-updates message just nod along

they're working on it so it won't be that long now

[violet heron]

Why are you linking fake game hacks

@viscid bramble UVR

[viscid bramble]

thank you

[magic badge]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[magic badge]

-rvc

[azure marsh]

Documentation

AI HUB Docs

https://docs.aihub.wtf/

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

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

[sudden wave]

uhh how do i paste a google drive link on illaria

it only allows me to drop hugging face links

but the voice model i have is only the zip file on google drive

i tried loading the zip file but it says error

:,,v

ayo no one is online

[zealous shuttle]

guess i use my original setup for now (mangio crepe for training, and rmvpe for inference) maybe adding titan for pretrain

[violet heron]

Load the .pth and .index that are inside the zip file

@sudden wave

[zealous shuttle]

it's been a while since that announcement

[thin stump]

just in time #🔥│model-maker-chat message

[sudden wave]

thanksss

[crimson umbra]

does anyone know why I cant switch the D0 Det thing to Crepe

[rare goblet]

Ayo? @crimson umbra level 1 !!!

[violet heron]

RMVPE is better anyways

[crimson umbra]

nah crepe was always better for me

[violet heron]

https://cdn.discordapp.com/emojis/1241507695744258048.webp?size=48

Crepe is way more sensitive and takes more GPU power compared to RMVPE

which also means a larger delay

[crimson umbra]

oh

ok

also

I dont think it's working

cause I have it set up so I can here it

and it wont work

[crimson umbra]

no not false

this one sounds like ass

no offenese

[violet heron]

Maybe it’s because your using RMVPE_onnx on a PyTorch model?

[crimson umbra]

and btw

[violet heron]

?

[crimson umbra]

nvm

[crimson umbra]

whats a pytorch?

[violet heron]

Framework most ai applications use

That’s where the .pth name comes from

[crimson umbra]

it's working now nvm

[violet heron]

https://cdn.discordapp.com/emojis/1258108640670384279.webp?size=48

[brittle pelican]

My friend pc is GeForce RTX 3060 Laptop GPU. Why cant open Applio ? :'0

[knotty moth]

update the gpu driver, and install CUDA toolkit perhaps

[brittle pelican]

How to check cuda?

:'0

[brittle pelican]

my friend was updated the driver

[midnight sonnet]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[brittle wing]

A gtx 1650 4gb ventus it s enough to train a rcv??????

[clever summit]

uh may i ask what "NameError: name 'gpus' is not defined" is. i didn't get this error when i used colab before

[brittle wing]

@violet heron

[clever summit]

nevermind i found my issue oh my goodness gracious

[rare goblet]

Ayo? @clever summit level 1 !!!

[clever summit]

actually i'm having issues with dependencies

[low shard]

no that's not powerful enough to train locally an RVC model, i suggest you to look up the google colabs cloud ways https://docs.aihub.wtf/essentials/how-to-make-voice-models/

[low shard]

what google colab are you using? send the link

[glad zealot]

that doesnt look like mainline so probably disconnected

[low shard]

i just checked rvc disconnected setup code and you are actually right lmao

lemme run it

[glad zealot]

im scares of that thing cause it updates packages

that could break stuff in the future

[low shard]

yea

i just run it and it works all fine for me tho

[clever summit]

when i first ran it, it ran fine

but then i got disconnected and had to rerun it and set everything up again

and now it's breaking

[glad zealot]

you could try other training colabs

inset self advert

[low shard]

Runtime > disconnect and delete runtime > refresh page, and re run everything could maybe work

[clever summit]

true....

[dawn tendon]

I need some help

So can someone help me?

[glad zealot]

help with what? cant really try to help without knowing what you need help with

[dawn tendon]

Last time i did that I didn't see any Url public link or something

Did i do something wrong?

[clever summit]

alright that did it

thank you

[glad zealot]

nope just press the arrow down to put it back to normal

[dawn tendon]

Ok

[clever summit]

nvm

[knotty moth]

check if all previous steps have run without issues and the dataset has loaded properly (not being unusually too fast)

[rotund prawn]

Hi

Can I ask you something?

[glad zealot]

sup

[rotund prawn]

Sup

How should we choose the target sample rate?

Should I choose according to the khz of my dataset?

[glad zealot]

yup

generally most of datasets are 32k

[rotund prawn]

Yes yes

I mean

I use a program(spek)

All my datasets appear to be 16khz

So

[glad zealot]

use 32 then

[rotund prawn]

16x2=32

Okeyyy

Thank you so much

[patent trellis]

-help

[azure marsh]

Command has expired

[patent trellis]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[patent trellis]

-colab

[azure marsh]

☁️ Google Colabs

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

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

[rotund prawn]

-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

Credits to Faze Masta and Antasma for compiling these links.

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

[maiden mesa]

So I’m using the weights bot to try to make an AI cover and when I press the model, it keeps saying doesn’t look like an audio file

[flint reef]

hey guys how do i get the index file for a model i only have the pth but not the index

i found the pth file online in a list but i dont know how i can get the index

[tame mica]

you don't really need an index

[flint reef]

it says upload failed for me

whenever i try to upload without an index

[tame mica]

what fork

[flint reef]

nvm fixed

[rare goblet]

Ayo? @flint reef level 1 !!!

[flint reef]

thx

[maiden mesa]

Ugh

[rare goblet]

Ayo? @maiden mesa level 1 !!!

[maiden mesa]

I need help