✨│ai-help

[craggy bramble]

just use chromium for Okada RVC so you don't have to run all the trash that chrome run in the background

[weary pond]

okay! Thank you Sushi, very helpful👍

[vague sapphire]

if using collab i could make it faster if i pay 10 bucks?

using a100 or v100 right?

[molten pecan]

Yes

Though if you have a fast PC you don't have to pay

If you use RVC locally

[vague sapphire]

amd = does not work correct?

[molten pecan]

I think so

[vague sapphire]

in free tier collab t4 is same as paid t4?

[velvet robin]

I installed what I need to get rid of the error but it's still there.

[craggy bramble]

I was thinking about testing if AMD training could be possible since people have done training on Intel ARC gpus but I replaced my server PC AMD gpu for NVIDIA.

If only I had an AMD and Intel I could possible make a guide for both of them but I don't make guides unless I've tested it myself

[vague sapphire]

i think it is possible but with my knowledge i am not sure how to set it up.
ive read that cuda doesnt work with amd but there is an alternative which i am not sure how to do

[sand iris]

If only okada had the brains

https://cdn.discordapp.com/emojis/1104640088840876093.webp?size=48&name=clueless&quality=lossless

[craggy bramble]

AMD gpus don't work well with Okada RVC either I had an AMD 6700 that could only run RVC at 192 chunks while my Nvidia 3060 12gb can run at 40 chunks

[sand iris]

Literally beyond stupid that "he doesn't have AMD hardware to test on"

DirectML is DX12 ffs !!!!

Also why the fuck do they bundle pytorch-cuda on the directml builds ? Why the fuck

[craggy bramble]

While the AMD 6700 is better than the Nvidia 3060 12gb for gaming

[craggy bramble]

maybe I'll buy an AMD and Intel for enthusiast sake since I really wanna buy a open case PC test bench literally just welded aluminum bars with a motherboard screw holes

[sand iris]

¯_(ツ)_/¯

[velvet robin]

I installed the required item but the error still persists.

[viscid moss]

[velvet robin]

Now I just need to know how to open RVC-TTS-WebUI.

[grand dirge]

can someone help me find the right settings for the voice changer???

[velvet robin]

How is that going to work?

[desert linden]

I did

It still just shows as blank

[viscid moss]

-rt

[azure marsh]

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

[viscid moss]

I'm not sure 😅

[velvet robin]

Well, who else does?

Don't leave me hanging.

[ember crow]

you're running a file called "app.py" in a virtual enviornment aka venv

idk what the purpose of venvs are btw

[viscid moss]

Try asking to I ned slep.

[sand iris]

Instead of installing python shit globally, and the python command refering to the "main" python install, you shadow the command to refer to any version, and have the packages installed elsewhere

Allows you not to break python packages when changing python version since the venv will remain on the old one

And also allows you to have different package configurations in different venvs, when a single install can't possibly handle everything

[vague sapphire]

can someone explain those

and these as well?

[ember crow]

what rvc webUI are you using

[vague sapphire]

tts

[ember crow]

ilaria tts?

[vague sapphire]

https://git.ecker.tech/mrq/ai-voice-cloning

this one

tortiose

i managed to get it to work but now i am just testing my model

i waas wondering what those did coz sometimes voice is good and bad depending on settingsa

[azure marsh]

Documentation

All-In-One English documentation

https://docs.aihub.wtf/

English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

More guides

How to Make AI Covers using Ilaria RVC

Link: Documentation

Create a model with RVC disconnected (colab)

Link: Google Docs

Credits: Nick088, Angetyde and Kanav

[timber canopy]

how do i add model in rvc

[vague sapphire]

are you working on colllab

[timber canopy]

i dont believe so

[vague sapphire]

then i wont be able to support you

sorry

[timber canopy]

how do i get the collab thing

[vague sapphire]

are you training a model or using one?

[timber canopy]

using one

[rare goblet]

Ayo? @timber canopy level 1 !!!

[proper shale]

#1159513888199540817 can help

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[vague sapphire]

Can my model be as good as 11labs?

[proper shale]

yes, if u have the right dataset

[soft stratus]

@proper shale how i make Crunch portguess talk in RVC

[proper shale]

?

Don't really understand what you mean

[soft stratus]

@proper shale i mean this

[proper shale]

just click on convert and it should work

[soft stratus]

Like original

I mean

@proper shale i tried with a pitch i don't know

@proper shale i mean to talk like original

[proper shale]

Try other values, like -6, -7

If you want it like the original then 0

[soft stratus]

@proper shale i tried 0 but not care

@proper shale i mean CTR Nitro Fueled in portguess talk in RVC like original

[proper shale]

Test some lower or higher transpose values depending on the ElevenLabs sample and the voice you want it to replicate

[glad zealot]

Lmao spam

[proper shale]

Like if the ElevenLabs voice is higher than the voice you want then transpose should be negative

[soft stratus]

Like emo

[proper shale]

If it's lower then increase

Just experiment with values

[soft stratus]

If i use the emotion on Elevenlab will succeed to RVC

[proper shale]

Yes

[hollow hare]

i cant use the ai voice because ever when i try use on discord or any game this ever cutting my voice

[proper shale]

increase chunks to like 320

so u prevent lag even when running stuff

[hollow hare]

already in every chunks

btw have any specifc download to 1660 super?

[unreal lichen]

hello how can i upload my model to this ??https://huggingface.co/spaces/r3gm/Advanced-RVC-Inference

[violet heron]

Model Downloader

[clever seal]

I'm trying to restart the space in Ilaria RVC, but can't find the 'settings' button. Was it moved?

[vague sapphire]

Cuda out of memory

What do I do

Empty.cache()??

[violet heron]

Under under the 3 dots

Only if it’s your space (so if you duplicated it)

[clever seal]

I gotcha. Thanks!

[clever patio]

How how do I run the local original RVC download
anyone know ?

[violet heron]

Extract everything and run the bat file

[hot lagoon]

for Applio-RVC-Fork wich UVR modles shoud I use?

[glad zealot]

ft

iirc theres one there that have "FT" in the nane

[hot lagoon]

just ft?

[rare goblet]

Ayo? @hot lagoon level 8 !!!

[glad zealot]

thats what i use so ig?

then also try mdx3 or whatever its called

[slow hill]

Rvc music links please

[brittle wing]

when i open the application nothin happens

[violet heron]

What file are you opening?

[brittle wing]

start_http

[vital mortar]

is there a local rvc web ui that has a way to record your own voice? im using the mangio_rvc

[rare goblet]

Ayo? @vital mortar level 2 !!!

[molten pecan]

Record it in real time?

For voice calls?

[ember crow]

try “go_web.bat”

[vital mortar]

no not the real time voice changer, the easy gui that i remember using in the google colab had one where you just record your voice, convert it and you can download it

[ember crow]

isn’t that just rvc

[molten pecan]

Oh so you want to do inference with an audio sample of your voice

Well for that you can just record yourself talking with any recording software

And save that recording and upload it and that's it

[vital mortar]

this is the one i have

i want to use this one with the easy quick record

[ember crow]

what colab are you using

[vital mortar]

Mangio-RVC-v23.7.0_INFER_TRAIN

locally

[ember crow]

then why did you say colab

[molten pecan]

-colab

[azure marsh]

All working colabs / spaces ☁️

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

Need some help? 🤔

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

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

[vital mortar]

because the one in the colab is the only one i knew it from

[ember crow]

oh

[vital mortar]

the EASY GUI but for LOCAL

[ember crow]

i can find that

[molten pecan]

Hey don't scream is midnight here

Have you tried googling it?

I'm sure it's pretty easy to find

[ember crow]

hang on

[vital mortar]

yeah but its not reliable and i know theres some people here with some good info you cant simply search

[ember crow]

h

can’t find it anywhere

[clever patio]

How did u get that one can u send the link

All these ones are shit the old AI hub one was the best and easiest

the google collab one

[vital mortar]

Mangio-RVC-Fork, by Mangio621
https://huggingface.co/MangioRVC/Mangio-RVC-Huggingface/tree/main

[clever patio]

ty

[clever patio]

Can anyone run 1 conversion for me nothing is working 😦

[vital mortar]

yea

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[rare goblet]

Ayo? @zenith yew level 4 !!!

[silver anvil]

yes, now explain your problem and what you're trying to do to someone there

@zenith yew

[zenith yew]

I'm trying to train a model voice in voicemy.ai, and it keeps on showing that message to me

I checked my pricing and I still have 0/1 out of training models

I was trying to do that export trick

[vague sapphire]

What if these two voices in an audio recording

[hot lagoon]

is there any other then uvr that can remove vocals?

[open sigil]

To remove vocals you can use this-https://colab.research.google.com/drive/1NA2GuJ2y-zRfoG3NearNiMCQa8NbidSh?usp=sharing

[open sigil]

You guys are not using gradio one?

[ember crow]

because webui is banned for normal users

[hot lagoon]

dose this use mdx modles?

[open sigil]

@frigid sapphire He might help with that.Not sure of these using mdx models or not.

[rare goblet]

Ayo? @open sigil level 1 !!!

[glad zealot]

yes and no

yes but some gatekept are working

[frigid sapphire]

aye - I am the creator of that notebook. This uses a cascaded pipeline with DEMUCS which is just another name for UVR followed by a Pyannote diarization.

The issue with raw UVE output is that it might still contain multiple speakers and non verbal cues like emotions, canned laughters, applauses etc. My pipeline gets rid of that completely.

[frigid sapphire]

if you have a credit card and a gmail ID, you can still use gradio webuis.

penned a piece here: https://black.dubverse.ai/p/running-rvc-models-on-the-easy-gui

[open sigil]

Any elevenlabs or RVC substitutes you know, that has better output?

[hot lagoon]

on mvsep shoud I use the one call Ensemble 4 models (vocals, instrum)?

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[lavish viper]

Pls help
https://cdn.discordapp.com/attachments/630882547777273868/1177226863781888041/image.png?ex=6571bcf4&is=655f47f4&hm=c99699e24dbd11e4195f0fff13ced26362f47d2279fe06ba28084de7628e5dcd&
not vorking :c

[rare goblet]

Ayo? @lavish viper level 1 !!!

[spiral osprey]

Does it work to cover music? or just train the model?

[molten pecan]

@low shard

[molten pecan]

Yes, it's for inference too

(AI covers)

[outer isle]

@molten pecan Are you here?

[molten pecan]

Yes

[worn narwhal]

Tyler Durden

pls

give me pls Tyler Durden

[tame mica]

no

[molten pecan]

Search on #1175430844685484042 or the website weights.gg

[rare goblet]

Ayo? @molten pecan level 47 !!!

[molten pecan]

Good luck!

[molten pecan]

Jesus fucking christ

[rare goblet]

Vote Booster: 10% (11 hours remaining)
✨ Tip: Use /card to modify your rank card

Vote Booster: Vote now for a 10% boost. https://arcane.bot/vote

[molten pecan]

@proper shale coming after you

:3

[rare goblet]

Vote Booster: Vote now for a 10% boost. https://arcane.bot/vote
✨ Tip: Use /card to modify your rank card

[tame mica]

uh oh

[copper pebble]

@modern oxide u on bro???

nvm

[modern oxide]

What?

[tame mica]

lel

[copper pebble]

can u link me the easy colab (training)

just want to test smt

[modern oxide]

https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw here ya go bbg

[copper pebble]

ty bro

[proper shale]

cant change batch size...

[low shard]

idk abt local bro my pc can run at max uvr 😭

[outer isle]

I have Problem with RVC, it takes too long with crepe

[molten pecan]

That's normal

Crepe takes a long time

But the results are worth it

Are you running it locally?

[proper shale]

no 💀

rvmpe is

faster and has 80% of the quality

[molten pecan]

70% of the time crepe sounds better to me 😭

I rather use crepe even if the quality is not a night and day difference compared to rmvpe

[robust vessel]

Thank you so much again for the tips. My model improved tremendously!!!

[spiral osprey]

what link to do the cover? Does Google Drive still work?

[copper pebble]

youre welcome

[molten pecan]

Yeah it still works

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[copper pebble]

how r u Julia didnt see you here for a while

[molten pecan]

Haven't seen me in a while?

I have been here yesterday actually

I'm doing good

Thanks for asking

[copper pebble]

my bad then

[outer isle]

Yes, I am running it locally

[molten pecan]

Do you have a really fast computer?

[outer isle]

but iNo, I'm Running Windows 10

[proper shale]

okay but

what GPU do you have

[outer isle]

Running on Intel devices

[molten pecan]

Which GPU?

[outer isle]

Intel

[proper shale]

Integrated?

HD Graphics?

[molten pecan]

Intel

[proper shale]

do you have any idea of how LITTLE that narrows it down

[vague sapphire]

What if there two voices in a wav file for training is there a way to separate those

Also is model.pth the only file needed for infernce and tts?

[outer isle]

Intel HD Graphics

[molten pecan]

Oh it's going to take a long time then

It's best if you use the huggingface spaces

Use Ilaria RVC

There's a guide on #1159513888199540817

[proper shale]

U can do that manually

[proper shale]

Yeah, but index is recommended

[vague sapphire]

How

[outer isle]

Oh! Snap

[proper shale]

just trim out those parts in audacity?

should be pretty easy but takes a bit of time

[vague sapphire]

I see

I thought there was a tool

[proper shale]

There was but it's not working anymore currently

[vague sapphire]

I see that’s understandable

[outer isle]

Can I have a link to Illaria RVC? @molten pecan

[molten pecan]

https://discord.com/channels/1159260121998827560/1167844377046040769

[outer isle]

Thank you

[outer isle]

I got an error where downloading the model

[molten pecan]

Are you logged in huggingface?

Do it if you are not

[molten pecan]

Is the link from huggingface?

[outer isle]

Yes, and I logged on too

[tame mica]

maybe you downloaded it with the same name as an existing model downloaded

[outer isle]

I don't have the file downloaded from Weights.gg

[tame mica]

i know

[half cove]

weights.gg models doesnt work

you need to reupload it to huggingface

or google drive

[carmine osprey]

how much usage hours we get in colab pro?

[summer dagger]

How I make voice models ? (:

[violet heron]

Wrong link

[violet heron]

Colab pro isn’t worth it

Just use alt accounts

[summer dagger]

Mister

[silver anvil]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Suggestions for @summer dagger

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[summer dagger]

What ?

[leaden sigil]

There's a link to the English how to guide on hwk to make a model.

[leaden sigil]

https://docs.aihub.wtf/

[vague sapphire]

do i need nvdia to use replay?

[rare goblet]

Ayo? @vague sapphire level 4 !!!

[summer dagger]

O

[leaden sigil]

If you have a good cpu, you can use that instead, but it is slower than Nvidia gpu.

If you have an Intel or AMD, you can instead use Apollio rvc.

[violet heron]

Don’t recommend applio

It’s a pain to install and stole code

[half cove]

ayy you got qc back

[vague sapphire]

what do they want me to input?

the path to my audio files?

[leaden sigil]

It all worked out.

[half cove]

[vague sapphire]

okay i think thats what was wante

d

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[vague sapphire]

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

how to setup tensorboard on this ?

[versed kite]

@violet heron hey, you linked me that RVC - can i upload all of my speech as one file? i looked through the #1159513888199540817 but there's different versions of this shit and it's more just a collection of links for the programs than tutorials

[violet heron]

As long as it’s clear and it only has the person who you want to talk talking

[versed kite]

dope, thank you

[violet heron]

Make sure there’s no long pauses

I would train using RVMPE

[versed kite]

i'm going off of a tutorial for a similar RVC interface, dude is using colab but i think there's enough similarity to get me through it

[violet heron]

Ok

[vague sapphire]

what abour using harvest ?

how good is that one?

[violet heron]

Really slow, RVMPE is faster and sounds better

[vague sapphire]

anyone have the link to the tensorboard page?

[vague sapphire]

thanks

anyone what this means?

[vague sapphire]

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

how to setup tensorboard on this ?

anyone know where the model is saved?

[brittle wing]

Where do I use the /cover bot thing?

[violet heron]

#1159290666019471421

Use wordband not astra labs

[violet heron]

pth file is saved in weights

Index is saved in logs

[vague sapphire]

can i use the ones with the e?

are those the previous epochs?

also do we change any of this?

also transpose

the audio i get sounds like the sample

nvm something is workin

[violet heron]

Yes

[vague sapphire]

can i use those as well ?

[violet heron]

Yes

[vague sapphire]

oh

[violet heron]

But I would use the one just with the model name

Since that has all epochs

[vague sapphire]

model.pth

okay

could i use this model for tts?

[violet heron]

You also need the index

Sketchy

@proven hill weird file

[proper shale]

Timing out

[proven hill]

!mute @brittle wing 1d suspicious file

[patent trellis]

@file.bekcisi muted

Reason: suspicious file
Duration: 23 hours, 59 minutes and 59 seconds

[vague sapphire]

there is only one index right?

[proper shale]

if its the added one u should be fine

[vague sapphire]

even with all different epochs?

[proper shale]

yup

[violet heron]

Yep

[vague sapphire]

is there a guide about index rate, vloume norm and consonant protection?

[proper shale]

not really

[violet heron]

Index rate controls how strong you want the index to be

[proper shale]

but index rate just uses more of the accent simulation and uses more of ur model as a result

[violet heron]

The index basically controls accent

[vague sapphire]

higher index means sound more like the input audio?

[proper shale]

more like the model

but keeping it at 1 can cause problems

[violet heron]

More stronger accent

[vague sapphire]

i see

[violet heron]

I use 0.4 - 0.66

[vague sapphire]

higher is more like the model.

what about volume norm and consonats protection?

to prevent slurs?'

[atomic mulch]

how do you smoothen out a voice to make it less robotic

[proper shale]

if increased, volume norm uses more of the model's volume, less uses the same input audio's volume

[vague sapphire]

ohh i see i understnad

[proper shale]

consonant protection is basically to prevent shit from breaking in certain input audios

normally dont need it

[brittle wing]

Thanks

[violet heron]

Use clean vocals without slurs or manually cut them out in a video/audio editor

[proper shale]

how does it actually sound? send a sample

[proper shale]

bro...

they didnt mean that

[vague sapphire]

what if the audio is yelling?

[rare goblet]

Ayo? @atomic mulch level 4 !!!

[proper shale]

they just asked if consonant protection was preventing those thinfs

[violet heron]

Head empty

[copper pebble]

Timbo

[proper shale]

This sounds pretty okay tbh

Can't quite hear the problem

[atomic mulch]

I need to learn these settings better

[proper shale]

...not really

[atomic mulch]

Do you mind dumbing them down?

[proper shale]

they dont really make that much of a difference but sure

resample messes with the sample rate of your audio

if u need it to be a specific sample rate, there's that

[atomic mulch]

what if u dont know what samplerate u need?

[proper shale]

just keep it at 0

[atomic mulch]

ok easy enough

thanks!

[proper shale]

it'll export to the max quality possivle

[atomic mulch]

at 0? (To confirm)

[proper shale]

yeah

[atomic mulch]

Okay thats ez to understand thanks! and what about the other 2?

[proper shale]

volume envelope basically mixes the volume of the input with the model's volume

less makes it more like the input audio

and voiceless consonants protection helps to prevent tearing and shit, but it reduces indexing accuracy because it messes with the sounds that index takes care of

aka the accents and stuff

[atomic mulch]

i see

so lowering the value makes it more accurate to the sample you provide, but at the cost of reducing accents of the model

[proper shale]

normally those settings wont make a big impact, so i wouldnt really mess with them a lot

id mess with indexing and maybe the voiceless consonants thing

[proper shale]

it protects the voiceless consonants and reduces the accuracy yeah

[atomic mulch]

is that what this is

[proper shale]

yup

[atomic mulch]

what does this slider do exactly

[proper shale]

set it to the lowest and it'll use more of the original input's mannerisms, and increasing it takes the mannerisms of your model to the output

set it to 1 and it'll replace all of it to sound like your model

[atomic mulch]

this is SUPER helpful

@proper shale what other setting do u recommend testing?

this one?

[proper shale]

pitch extraction, transposing, and that one yeah

[atomic mulch]

isnt rmvpe the best always?

[proper shale]

ehhh yes... and no

80% of the time, yes, rvmpe is better

but mangio-crepe has a lil bit more quality and less artifacting in low hop-sizes. However, mangio-crepe takes more time in lower hop-sizes as it's using a shorter part of the audio to analyze pitch

and

mangio tends to not like noisy stuff

very noise sensitive

[atomic mulch]

so with an accurate base sample, and with good hardware, low hop mangio crepe is the best?

[proper shale]

yeah, but i definitely suggest using both

rvmpe and crepe

because they are excellent

[atomic mulch]

regular crepe or mangio-crepe

[proper shale]

mangio

[atomic mulch]

dude you are amazing

[proper shale]

aw thank you

[atomic mulch]

seriously ive been struggling for so long and u made my day

[half cove]

common mj w

[proper shale]

that's epic to hear

[atomic mulch]

No, thank yyou!

I'm a content creator who makes animations and i really needed to get this done for a video. You helped so much with understanding this

[proper shale]

Yoooo, nicee

Glad to have helped :)

[granite yarrow]

hi, I need help while starting the voice-changer, when i open the app iit appears a message that says wait web server... 10, 20, 30... and then it continues infinitly, how do I resolve that?

[proper shale]

Do you have an antivirus/firewall active?

If so, deactivate them to run W-Okada, as the antiviruses can mess with the local web server hosting

[granite yarrow]

the windous antivirus too?

[rare goblet]

Ayo? @granite yarrow level 1 !!!

[proper shale]

There is a possibility that this can be interfering with it, yes

[atomic mulch]

What does this setting change?

[proper shale]

applies median filtering, which basically makes everything "average"

this helps to average out stuff like breathing

as the little thing says there

[granite yarrow]

I think that it doesnt was the antivirus

[atomic mulch]

isnt that kinda a bad thing?

[proper shale]

yeah it can make the quality somewhat worse

[atomic mulch]

so to disable it make it 2 or below?

[proper shale]

Try deleting stored_setting.json and restarting the voice changer

[proper shale]

Yup

[atomic mulch]

do u find urself doing that alot?

[proper shale]

but for the most part... not really

[proper shale]

honestly i just increase it when i wanna see different results

[atomic mulch]

so u only do 3 and above?

[proper shale]

yeh

[granite yarrow]

eitaaaa

obg irmao

[median monolith]

The weights.gg model link is not working. Am i using the wrong link?

[lavish viper]

Hello, I would like to ask you for help
I do everything as described but only errors appear in the console :c

[oblique rain]

you don't have to paste weights.gg's link, when you select your model, tap on the three-dot icon and click copy model link

[uneven holly]

how to use rvc v2 disconnected

[rare goblet]

Ayo? @uneven holly level 2 !!!

[hollow torrent]

How much better is Applio better than mangio?

[rare goblet]

Ayo? @hollow torrent level 1 !!!

[median monolith]

It also dint work

[proper shale]

de nada ;)

[proper shale]

it should be the download link

[ruby marlin]

how do I get discord to pick up the audio from the app?

[proper shale]

oh maybe try uploading to drive

[proper shale]

virtual audio cable

[ruby marlin]

a what now?

[proper shale]

it actually isn't that better, just has some useful quality of life features

[uneven holly]

.

[proper shale]

but at the same time brings way too many hassles

[proper shale]

#1159513888199540817

[proper shale]

A virtual audio cable - it transmits an audio from one app to another.

[ruby marlin]

I think I have one of those (VoiceMeeter) but idk how to tell it to use the voice changer client

[proper shale]

make sure the output in W-Okada is voicemeeter and the input in discord is voicemeeter

so it recognizes the ai voice as ur mic, basically.

[ruby marlin]

kk

[rare goblet]

Ayo? @ruby marlin level 1 !!!

[ruby marlin]

it seems to be working, the audio is just glitching out

i gtg. ill fix that later

[proper shale]

alrighty

[proper shale]

you need to apply kalo's local fixes, i think its in the original mangio guide

[lavish viper]

Where can I find this guide?

[proper shale]

no, this

https://github.com/kalomaze/Mangio-Kalo-Tweaks/releases/tag/v3.3

@desert linden just found out what we were doing wrong

you needed to write in terminal

run run.sh

wait lemme get the right command here

[proper shale]

In this terminal, type ‘cd’ with a space, and then drag and drop the Mangio RVC folder on top of the terminal text window to complete the path. Hit enter. Your path should change to the Mangio folder one. Then, you can type ‘sh run.sh’ in the terminal and hit enter. This will start the first time installation process. Wait for it to complete installing everything; after this is done, it should launch the RVC web ui properly and work from here on out if you launch it in the same way.

That's why we were doing it wrong

[proper shale]

get the my_utils.py from the mangio kalo tweaks

and replace it

[vague sapphire]

anyone know why it disconnects from collab ?

[rare goblet]

Ayo? @vague sapphire level 5 !!!

[proper shale]

after 4 hours it does that

[vague sapphire]

even if i reconnect it disconnects when making a wav file

is there a size limit to the audio file i can use for infernce?

im using the collab file that is easy 10-25

[proper shale]

oh that

easy guy

easyGui

[vague sapphire]

no gui though

[proper shale]

it is

because theres a guy emoji

guy, gui

get it?

[vague sapphire]

ohhh

i get it

[proper shale]

anyways I'd suggest using other stuff, like RVC-HF

in #1159949278270193734

[vague sapphire]

okay

[proper shale]

name has spaces

[gloomy surge]

From the enter the audio file section, I click on the browse section. Since I need to select mp3 files, I have to select mp3, but I do not have any mp3 files.

[violet heron]

No space in mode name

[violet heron]

Wav and FLAC files work as well

[lavish viper]

this is how it works if I download it manually, where should I copy it?

[rare goblet]

Ayo? @lavish viper level 2 !!!

[gloomy surge]

ok thank you

[lavish viper]

or should I close it and open it again?

[oblique rain]

I guess weights links are not supported

[hybrid spire]

this is a totally separate qs, just outta sheer curiosity, what's w/ the huge number of Polish voice models? I thought this RVC software predominantly was built around English language

its interesting to me to see that the popular character models are also popping up under dif languages

[violet heron]

RVC works with most languages

@shy zealot categorized a large amount of models as the wrong language

[hybrid spire]

Neat.

Plus, I never thought that I'd see so many for Polish specifically. I thought it'd be more from like the frequently recognized languages, like Spanish or French

[median monolith]

Should I then delete something from that link in that image to make It work? In anycase, All of the weight.gg model links as i have seen are like this, so if they dont work in someway, then i guess all of the weights.gg models are only usable in illaria rvc if you upload them to google drive.

[rare goblet]

Ayo? @median monolith level 2 !!!

[olive olive]

Guys why when i train my model locally i have no .pth file in weights ?

i tried many time and it won't come

everything good i have index but never .pth

[spiral osprey]

does the google drive link still work in easy gui?

[rare goblet]

Ayo? @spiral osprey level 5 !!!

[half cove]

i think

[spiral osprey]

Which haggingface license does it have to be? Was it openrail?

[proper shale]

Probably unchecked the save small finished model tick

[half cove]

yep

[vague sapphire]

if im using rvc v2 disconnected and if i alrady have a model do i only export model from drive?

[dull wave]

how to so vits svc remodel to rvc?

[half cove]

[iron swan]

in wokada whenever i try to add a new model the file explorer wont respond, something got a solution to this?, thanks

[molten pecan]

@violet heron ogm 😭😭😭

-svc

[azure marsh]

Yeah I still use so-vits-SVC

https://tenor.com/view/crowd-avoiding-man-gif-25257570

[molten pecan]

#🔍│help-w-okada

[dull wave]

what?

[rare goblet]

Ayo? @dull wave level 1 !!!

[violet heron]

-svc

[azure marsh]

Yeah I still use so-vits-SVC

https://tenor.com/view/crowd-avoiding-man-gif-25257570

[noble heath]

I need a link for rvc

cans someone pls give me the link

[tropic spoke]

Can someone help me with this issue. I was trying to separate the vocals of a song on RVC UI.

[marsh grove]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Suggestions for @noble heath

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[molten pecan]

Change the name of the file

To something like "test"

It seems like the app doesn't support other foreign letters

[vague sapphire]

-svc

[azure marsh]

Yeah I still use so-vits-SVC

https://tenor.com/view/crowd-avoiding-man-gif-25257570

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[vague sapphire]

🥲

[tropic spoke]

I just tried that and got the same error

[radiant flare]

sorry if the is the wrong spot
im using huggingface.co, my audio file processed... i can hit the play button to listen to it...

the download options fail

[rare goblet]

Ayo? @radiant flare level 1 !!!

[molten pecan]

Click the share button

It will open a tab

I think there should be a tab named File

Click on it and there's the audio

[radiant flare]

clicked it, its flashing

[molten pecan]

Can you send sc?

[radiant flare]

i can share my screen i guess

[molten pecan]

Maybe tap the buttons on the right and hit Download

[radiant flare]

the 3 dots download option tries / fails to download a html page

[molten pecan]

Dang

Yeah that happens sometimes

I'm afraid to say this but

I think you gotta do it again

Dont tap out the window this time

[little crown]

how can i change my voice?

[molten pecan]

That will probably fix it

[little crown]

pls help

[molten pecan]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Suggestions for @little crown

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[molten pecan]

-okada

-realtime

[little crown]

nonono

[azure marsh]

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

[little crown]

how do i use one

[molten pecan]

There's a guide there

[little crown]

ok thx

[molten pecan]

And plenty more on #1159513888199540817

Good luck!

[radiant flare]

right clicking on the dots did the trick

[vague sapphire]

for hina mod collab can i copy to my own drive and then run it ?

[proper shale]

no

[vague sapphire]

why?

[proper shale]

no wait

i thought u were referring to voice changer shit and syncing to drive

lolll

[vague sapphire]

i think this one is the audio converter

[proper shale]

i mean yeah sure, you can copy the thing to your own drive, but you'd lose the updates

[vague sapphire]

omg LUL

im such a big fool;

baka

[hoary elbow]

need some help, I can never successfully run a cover
here is the log

C:\RVC1006AMD_Intel>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897 --dml
2023-11-24 09:10:36 | INFO | configs.config | No supported Nvidia GPU found
overwrite preprocess and configs.json
2023-11-24 09:10:36 | INFO | configs.config | Use DirectML instead
is_half:False, device:privateuseone:0
2023-11-24 09:10:41 | INFO | __main__ | Use Language: en_US
Running on local URL:  http://0.0.0.0:7897
2023-11-24 09:11:13 | INFO | infer.modules.vc.modules | Get sid: youzhanv2-xi.pth
2023-11-24 09:11:13 | INFO | infer.modules.vc.modules | Loading: assets/weights/youzhanv2-xi.pth
2023-11-24 09:11:14 | INFO | infer.modules.vc.modules | Select index:
2023-11-24 09:11:16 | INFO | infer.modules.vc.modules | Get sid: nanachi_MangioCrepe.pth
2023-11-24 09:11:16 | INFO | infer.modules.vc.modules | Loading: assets/weights/nanachi_MangioCrepe.pth
2023-11-24 09:11:17 | INFO | infer.modules.vc.modules | Select index:
2023-11-24 09:12:46 | INFO | infer.modules.vc.pipeline | Loading rmvpe model,assets/rmvpe/rmvpe.pt
2023-11-24 09:12:48.1284232 [W:onnxruntime:, session_state.cc:1169 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf.
2023-11-24 09:12:48.1345560 [W:onnxruntime:, session_state.cc:1171 onnxruntime::VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments.

C:\RVC1006AMD_Intel>pause
Press any key to continue . . .

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI
here is the repo i'm using

[sand iris]

Been a real long while since I looked into it, but isn't there something about "converting models to onnx" in the UI ?

I might be completely wrong tough

[vague sapphire]

when training a model is it always english or any language?

[proper shale]

any language

[wheat chasm]

Can I get some help with my dataset? For some reason it sounds absolutely nothing like the character and I can't figure out why. It's not the best dataset, but I don't think it's so outrageously bad that it should sound the way it does

Am I allowed to post a dataset here if it's voice clips from an anime

It's all in one audio file

[primal remnant]

Having some issues with the applio rvc fork, was wondering if anyone experienced something similar and found a fix, i've tried some of the givens like reinstalling the program, restarting my pc, reinstalling fairseq, etc...

Traceback (most recent call last): File "C:\Users\\Desktop\Applio-RVC-Forked\Applio-RVC-Fork\infer-web.py", line 26, in <module> import fairseq File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\site-packages\fairseq\__init__.py", line 20, in <module> from fairseq.distributed import utils as distributed_utils File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\site-packages\fairseq\distributed\__init__.py", line 7, in <module> from .fully_sharded_data_parallel import ( File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\site-packages\fairseq\distributed\fully_sharded_data_parallel.py", line 10, in <module> from fairseq.dataclass.configs import DistributedTrainingConfig File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\site-packages\fairseq\dataclass\__init__.py", line 6, in <module> from .configs import FairseqDataclass File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\site-packages\fairseq\dataclass\configs.py", line 1104, in <module> @dataclass ^^^^^^^^^ File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\dataclasses.py", line 1230, in dataclass return wrap(cls) ^^^^^^^^^ File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\dataclasses.py", line 1220, in wrap return _process_class(cls, init, repr, eq, order, unsafe_hash, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\dataclasses.py", line 958, in _process_class cls_fields.append(_get_field(cls, name, type, kw_only)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\\AppData\Local\Programs\Python\Python311\Lib\dataclasses.py", line 815, in _get_field raise ValueError(f'mutable default {type(f.default)} for field ' ValueError: mutable default <class 'fairseq.dataclass.configs.CommonConfig'> for field common is not allowed: use default_factory

[primal remnant]

How does the model sound without the inference?

[tender briar]

C: \Users\ Beary \ OneDrive\ Documents\ Downloads\MMVCServerSIO_win_onngpu-cuda_v. 1.5.3.17\MMVCServerSIO>MMVCServerSIO.exe
-p 18888
--https false
-content_vec_500 pretrain/checkpoint_best_legacy_500.pt
--content_vec_500_onnx pre
train/content_vec_500.onnx
-content_vec_500_onx_on true
--hubert_base pretrain/hubert_base.pt
e_jp pretrain/rinna_hubert_base_jp.pt
--hubert_soft pretrain/hubert/hubert-soft-0d54a1f4.pt
--hubert_bas
--nsf_hifigan pretra
in/nsf_hifigan/model
--crepe_onnx_full pretrain/ crepe_onnx_full.onnx
--crepe_onnx_tiny pretrain/crepe_onx_tiny.
onnx
--rmvpe pretrain/rmvpe.pt
--model_dir model_dir
-samples samples. json
Traceback (most recent call last):
File
"MMVCServerSIO.py", line 33, in <module>
File "nods ihoo-control.y, line fio, " instinlize
File "logging_
", line 1169, in
File "logging_.
init
__•ру"
', line 1201, in _open
FileNotFoundError: [Errno 2] No such file or directory: 'C: \Users\ \Beary\OneDrive\Documents\Downloads\MMVCServerSIO _win_onnxgpu-cuda_v.1.5.3.17\MMVCServerSIO|vcclient.log'

w-okada voice changer

[wheat chasm]

Just like a regular voice I guess? Idk

[primal remnant]

If you go to the file directory does the file "MMVCServerSIO|vcclient.log" exist? if not you could try creating the file and see if that fixes your logging error.

[primal remnant]

So without the inference it also dosen't sound like the character?, Does the character have any kind of accent?

[wheat chasm]

Nah it's just like general american

[rare goblet]

Ayo? @wheat chasm level 1 !!!

[wheat chasm]

I can like DM you the audio I was using for the dataset, the audio I was using to test it, and how it came out if you want

[primal remnant]

Have you tried using the model with different vocals, it might be that the particular person speaks with such a high pitched voice or something it is struggling, and sure

[rose pond]

Anyone had experience running rvs not on google colab?
Downloaded Python and the 7z file but i'm kinda stuck

[violet heron]

Did you extract the 7z file?

[rose pond]

I tried, and then gave up after it was at 30 percent for almost half an hour

[violet heron]

Ok so it’s not RVC itself

It’s whatever program your using to extract it

[rose pond]

I think so,I just downloaded the 1st one on the microsoft store

[violet heron]

I would use 7zip

[rose pond]

Done 🙂 It's extracting

[shy rapids]

Where do I start?

I came here with no prior experience on what to do. 😭

[molten pecan]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Suggestions for @shy rapids

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[molten pecan]

Good luck!

[shy rapids]

thx

[rose pond]

I got it !! i think

I have no idea where I can upload a model into thing or where I can choose to use my own microphone instead of drop an audio file

[rose pond]

Update: Got all the index files and stuff for the models in the right place, but still getting error messages when I try to convert an audio file

[rose pond]

any advice?

[molten pecan]

What are you using?

Applio?

[rare goblet]

Ayo? @rose pond level 3 !!!

[rose pond]

Nvidia I think? lmao

[molten pecan]

I'm talking about the app

Are you using Colab?

Applio?

Gradio?

Huggingface?

kits.ai?

[rose pond]

🧍‍♂️
it says WebUI on the top?

[molten pecan]

Can you send a sc?

[rose pond]

mb forgot u could send screenshots

[molten pecan]

ahh interesting

At the bottom of the page it should say either Gradio or Applio

What does it say?

[rose pond]

Gradio

[molten pecan]

i see

I use the same :3

[rose pond]

ayy

[molten pecan]

Okay so

Have you tried deleting the text in the directory spaces?

Like in "enter the path of the audio file"

[rose pond]

my connection timed out lmaooo

one sec

[molten pecan]

[primal remnant]

`At this point i've tried reinstalling python and buildtools to no avail, even tried installing fairseq in editable mode, but I'm getting seperate issues there

File "C:\Users\ \AppData\Local\Temp\pip-build-env-0_ut8jng\overlay\Lib\site-packages\setuptools_distutils_msvccompiler.py", line 421, in compile
raise CompileError(msg)
distutils.errors.CompileError: command 'C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.38.33130\bin\HostX86\x64\cl.exe' failed with exit code 2
C:\Users\ \AppData\Local\Temp\pip-build-env-0_ut8jng\overlay\Lib\site-packages\setuptools_distutils\dist.py:988: _DebuggingTips: Problem in editable installation.
!!
!!
cmd_obj.run()
error: command 'C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Tools\MSVC\14.38.33130\bin\HostX86\x64\cl.exe' failed with exit code 2
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building editable for fairseq
Failed to build fairseq
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects

Did some research about this particular error and it seems like a few other people have experienced it, seems like the issue is generally unsolved besides a recommendation for installing the latest SDK for windows, tried updating my sdk for windows 10 to make sure it was up to date and that resolved nothing, really scratching my head here at this point.`

[rare goblet]

Ayo? @primal remnant level 3 !!!

[molten pecan]

https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC-beta.7z use this one then

It's a lil outdated but to me sounds the same quality-wise and settings-wise

It's super easy to download and you don't need to install anything

Just unzip it and open go-web and that's it

Applio is harder to install and the installation seems to be very unstable from what I seen

[rose pond]

im gonna try that

[molten pecan]

Wait what why

[rose pond]

huh

[molten pecan]

Oh 😭

Nevermind. Go ahead

[rose pond]

I cant seem to get my other one to work

[molten pecan]

Have you deleted the text?

[rose pond]

I deleted the text, but now it's saying there's an error with the volume envelope scaling

[molten pecan]

Try leaving it at 0

[primal remnant]

mm thanks, i'll just stick with the regular mangio fork for awhile longer I guess, see if I can't find another fix for this later.

[molten pecan]

Alright then

Good luck!

[rose pond]

the default was at .33, but neither leaving it or shifting down to 0 works

honestly at this point i'm thinking it might be something wrong w the model

[green birch]

i been trying to fix my model for 2 weeks but cant seem to get the help

no matter train time all turns out the same

[primal remnant]

gonna try one more thing right now, going to downgrade from python 3.11 to 3.9.8

[feral sphinx]

I'm trying to train AI Voice what to use ?

[rose pond]

Do I need to download 'Triton' to use Gradio

[stoic stump]

the colabs not working anymore ):

[primal remnant]

Finally fixed it, if anyone else has this issue, what I did to fix it was the following:

• Install Python 3.9.8
• Use the applio gui installer
• Place the non-gui installer version of applio in the folder that contains the dependencies.bat, install_Applio, install_Applio_amd
• Run the non-gui installer select option 1
• After install, run non-gui installer one more time with option 1
• Finally launch the go-applio.bat

[obtuse galleon]

im getting a runtime error every time that i try to duplicate a space on huggingface, anyone knows why?

[proud elbow]

there's "no module Triton" message in RVC0813 though I didn't notice it as a serious issue, anyway try download the latest version
https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006Nvidia.7z

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[zealous shuttle]

for those who use this colab, what is the best settings for it? i need reference

Its AICover NoUi

[copper pebble]

#1159513888199540817

[static gyro]

i have questions.
A series just came out and i want to do one of the voices for it.
would it be spoilers to just do it now or do i wait a week?

[copper pebble]

does it come out all at once or one episode each day?

[vapid jasper]

How much epochs should i put, if i have 15min voice record?

[rare goblet]

Ayo? @vapid jasper level 1 !!!

[ember crow]

what colab do you mean?

[brittle wing]

hako

help

it woont work

i got 1050 ti

[ember crow]

wdym by it

what fork are you using

[brittle wing]

???

[rare goblet]

Ayo? @brittle wing level 1 !!!

[ember crow]

are you using applio, mangio, or something else

[brittle wing]

???

[ember crow]

what isn't working

[brittle wing]

no voice it not recognizing im talking

[ember crow]

but what are you using

[brittle wing]

windows?

[ember crow]

not what os you're using

[brittle wing]

rvc voice changer

real time voice changer

[ember crow]

w-okada?

[brittle wing]

ion know

https://huggingface.co/wok000/vcclient000/tree/main'

wait

i found the problem

my headset was muted

can u help me with the settings?

@ember crow

[ember crow]

that's w-okada

[brittle wing]

or anyone

[sand iris]

https://rentry.co/VoiceChangerGuide

[brittle wing]

oh, so wrong channel>

[brittle wing]

thank u mi amor

[brittle wing]

hi, i changed the settings and now i magically cant hear things

i can only hear my self

(passthrough)

wtf

it just started working

nevermind bruh

sorry

[azure marsh]

Hey, @vapid jasper!

👇 Here are some resources to help you identify if your model is overtraining

How do I know if my model is overtaining?

All-In-One Guide on how to make a good model

This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[solemn aspen]

im having troubles setting up the vb audio cables to discord can anyone help?

[brittle wing]

💀

[lament eagle]

Lmfao who made this

[zealous shuttle]

i just rename it to prefen for banning

rename all the cell

[lament eagle]

The name is funny but I mean the content

couldn't you like mix it with FL

[jaunty nacelle]

hello everyone, i just installed RVC v2 on my pc, and I also installed all libraries from requirements.txt, however, I can't run the app, it gives me the following response in the log:

Traceback (most recent call last):
File "C:\Users\Admin\Desktop\RVC\RVC-GUI-main\rvcgui.py", line 23, in <module>
from vc_infer_pipeline import VC
File "C:\Users\Admin\Desktop\RVC\RVC-GUI-main\vc_infer_pipeline.py", line 1, in <module>
import numpy as np, parselmouth, torch, pdb
ModuleNotFoundError: No module named 'parselmouth'

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for fairseq
Failed to build fairseq
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects

[static gyro]

all at once

[woeful sedge]

in the noui colab for model inference, how do you rise the pitch of the vocals

[proper shale]

probably has an option named transpose

thats where u mess with

the pitch

[woeful sedge]

theres only

pitch change

and pitch change all

[proper shale]

yeah those can help

[woeful sedge]

but when i set it to 12

[rare goblet]

Ayo? @woeful sedge level 1 !!!

[woeful sedge]

it kind of got deeper

wait

nevermind

i just set it too high

[proper shale]

yeah

[woeful sedge]

what for male to female

ima try 5

6*

[proper shale]

i think its in octaves? im not sure

what does the colab say

[woeful sedge]

does anyone have patrick star model lil

[grizzled token]

Hi everyone. Can you tell me which part of the GPU is important to have powerful to be able to use the models in real time? VRAM or something else?

[proper shale]

VRAM and CUDA cores mainly

[grizzled token]

In thanks

[rare goblet]

Ayo? @grizzled token level 1 !!!

[tribal cosmos]

@proper shale

btw this is what it looks like

[proper shale]

oh yeah... keep at 10%

[tribal cosmos]

oh ok

[finite galleon]

Wait whaaa

[tribal cosmos]

?

[finite galleon]

Whats that

[tribal cosmos]

idk a thing called collapse

[proper shale]

to prevent mode collapses

model*

[finite galleon]

Dangg

Rvc disconnected needs to take notes innit

[tribal cosmos]

btw how do i check if my model overtrained?

[finite galleon]

I only user colabs to train so i dont know thats a thing

[proper shale]

it didn't overtrain yet

uhhhh im a bit worried about how many times youve restarted the training and returned to a lower epoch

[tribal cosmos]

yeah i had collapse turned all the way up

[tribal cosmos]

how do u tell, i wanna be able to tell myself so i dont gotta ask people everytime

[proper shale]

well theres a whole guide for it

-overtrain

[azure marsh]

How do I know if my model is overtaining?

All-In-One Guide on how to make a good model

This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[proper shale]

but uhh

tldr

g/total stops rising? check d/total. d total rising? thats a red flag

if not, keep going

[hot lagoon]

in uvr What dose Demucs model do?

[tribal cosmos]

whats tldr?

[proper shale]

too long, didnt read

[tribal cosmos]

where do i see dis?

[proper shale]

TensorBoard

[tribal cosmos]

like where on tensor board

[proper shale]

scalars

just type in "total"

and you'll see g and d

[tribal cosmos]

this is total d

[hot lagoon]

💀

[tribal cosmos]

what do i look for

[proper shale]

on the scalars tab

[tribal cosmos]

yeah i mean i already did that

but like what im lookin for to see if it overtraied

[proper shale]

yeah uh

if it starts rising and g/total also starts rising

[hot lagoon]

Wait did they add Thershold % for collapse for colab user?

[proper shale]

but what the hell happened with the training session 💀 did you restart it way too much?

the graph looks like spaghetti

[tribal cosmos]

bro some kid told me to do all the way

[tribal cosmos]

yeah imma retrain ill resend graph when i retrain

[proper shale]

its in applio i believe

[hot lagoon]

they just add this?

[tribal cosmos]

10 is good right?

[proper shale]

ye ok

[glad zealot]

im concerend about those grapth

[hot lagoon]

me to

💀

[tribal cosmos]

ngl the model turned out really good tho

[rare goblet]

Ayo? @tribal cosmos level 5 !!!

[tribal cosmos]

lemme see if i can find any stuff i did with it

[hot lagoon]

Also what is Thershold % for collapse ? Like I just now know about this is this like a new update today?

[proper shale]

na this is kinda old by this point

[tribal cosmos]

this is the crazy graph model

[proper shale]

this helps to prevent model collapses

[hot lagoon]

well what is a good % then for it ?

[tribal cosmos]

waiit then whats so bad about turning it up all the way, it just means more preventions right?

[proper shale]

not really

it could stop at something which isnt a mode collapse

[proper shale]

5, 10%? never tried the feature tbh

[tribal cosmos]

shit bro

u mean 10% not 10

cuz its 0-50

[lament eagle]

The fuck are you using lmfaooo

[rare goblet]

Ayo? @lament eagle level 13 !!!

[proper shale]

but it says like, % in the option

that's why im saying 10%

[tribal cosmos]

shitt ohh nvm so im good

[proper shale]

because it says threshold percentage

yh

[tribal cosmos]

secret sauce 😎

[lament eagle]

You using Applio?

[lament eagle]

Wtf is that graph though 😭

It's so fucking funny

@tribal cosmos Btw is that local Applio?

[proper shale]

yeah

[lament eagle]

I haven't checked colab Applio for training

I really wonder what threshold % for collapse does

and would it make the model better

[hot lagoon]

me to

[proper shale]

prevents model collapses

which make ur model stop improving like before

[lament eagle]

What does collapse mean in graph anyway

[proper shale]

so yeah it technically makes ur model "better"

[proper shale]

big drop in g/total

[lament eagle]

Model training guide just explains overtraining and it's not even detailed

[hot lagoon]

Right now haveing a "Modle block aka art block but for modles

[proper shale]

-overtrain

[azure marsh]

How do I know if my model is overtaining?

All-In-One Guide on how to make a good model

This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[proper shale]

theres this

[little crown]

how do i use the .zips

[brittle wing]

where can i download applio

where?

[proper shale]

#1159949278270193734

[proper shale]

#1159513888199540817

[brittle wing]

is applio better than python?

is it better

?

[glad zealot]

???

[lament eagle]

Damn nobody told me this

really useful

[brittle wing]

is applio better than python?

[proper shale]

wha

[brittle wing]

?

[proper shale]

python and applio are different things

applio uses python

python is a programming language, applio is a fork of RVC that uses said programming language

[brittle wing]

this chat is inactive not many people talk in here......

[lament eagle]

Bro just start a convo lol

nobody talks when there's nothing talk

[hot lagoon]

i check Applio colab it dose not see they have it the Thershold % for collapse.

[lament eagle]

Lol

guess I'm staying in rvcDisconnected

[tribal cosmos]

u gotta check this

[sturdy yacht]

does anyone here run on local install of rvc and have a nvidea RTX 3080 i need to see your cuda toolset setup

[tribal cosmos]

me bro

[vital field]

help

help

Please

[rare goblet]

Ayo? @vital field level 2 !!!

[distant pivot]

So I need help with this AI voice changer

After I recorded my voice I cant hear the AI one/how do I play it?

[vital field]

Please help

[tribal cosmos]

with what bro

u gotta be specific

[violet heron]

Did you click start

And what’s your GPU?

[distant pivot]

Yes I clicked start But I cant hear it. And tbf im not sure what my GPU is but I think its RT

[violet heron]

Go to task manager

Go to peformace

And click GPU 1

If there’s no GPU 1 click GPU 0

[distant pivot]

Alright I will try that

thank you

[violet heron]

In the top right is your GPU

[vital field]

It says qune

[violet heron]

Passthru on?

[vital field]

What do I do

[violet heron]

Dunno

[distant pivot]

I dont have task manager and yes pasthru on

[rare goblet]

Ayo? @distant pivot level 1 !!!

[violet heron]

Wait till a helper comes

[violet heron]

Comes preinstalled with windows

[latent kettle]

I'm getting this error while training index in RVC disconnect

[distant pivot]

hmmm

[crude grail]

I'm back... After getting a malware.

[violet heron]

Did you train the model yet?

[violet heron]

What did you download?!

[crude grail]

A miner

by accident

[violet heron]

What were you trying to download

[crude grail]

Man, I had to format

[distant pivot]

Somehow I dont have taskmanager

[latent kettle]

Just for testing I have trained a model on 10 epochs. But on index training I'm getting this error

[violet heron]

Restart your runtime and re-enter everything

[crude grail]

A working Portal 2. Because Steam was lazy to run it.

No lie

[violet heron]

Piracy Moment

[crude grail]

It opened and then closed

so

[crude grail]

yes.

[latent kettle]

I have to do everything again?

[crude grail]

Lesson learnt: don't download from ElAmigos

[rare goblet]

Ayo? @crude grail level 1 !!!

[violet heron]

Did you select the audio you want to convert

[latent kettle]

So I have to do everything again?

[violet heron]

What’s your GPU

[violet heron]

I don’t think you have to rerun the pre processing cells

[vital field]

0

[violet heron]

You don’t have a GPU?

[sturdy yacht]

i dmed u

[latent kettle]

From which cell I have to continue?

[violet heron]

Pitch extraction

And then train index again

[sturdy yacht]

i meant 3060 rtx 3060 *

[vital field]

Quadro P1000

[sturdy yacht]

AORUS 3060 having issues with cuda for some reason

[violet heron]

You can edit messages

[sturdy yacht]

yea sorry

just need someone with another 3060 with rvc locally installed and still working for them

[vital field]

thats my gpu

[violet heron]

Should probably work

I’m not sure

You can wait till a helper comes

Did you try closing it and reopening it?

[vital field]

Yes I did I restrart on my computer

[rare goblet]

Ayo? @vital field level 3 !!!

[sturdy yacht]

anyone here with a local install of rvc on GPU RTX 3060 still working for them ?

[rare goblet]

Ayo? @sturdy yacht level 4 !!!

[latent kettle]

Use RVC nvidia

[sturdy yacht]

i did but rvc thinks cuda is not available

[sturdy mountain]

what

how did u hacked

[crude grail]

Steam couldn't run portal 2 so

[crude grail]

Tried to find a working portal 2.

[latent kettle]

Can someone help me to solve this?

[astral jungle]

Can you scroll up a bit?

[sturdy mountain]

and u got hacked

[rare goblet]

Ayo? @sturdy mountain level 2 !!!

[crude grail]

got a cryptominer

[astral jungle]

Have you run this step yet?

[latent kettle]

Yes and I have also done with training of models. But I'm getting error on training index so I restart the runtime

[crude grail]

A question. I once tried to make a model, but when trying to do the final steps, shown me that the features weren't there or whatever