✨│ai-help

[brittle wing]

I've tried several different models and it persists

[wispy burrow]

What error? There could be several reasons.

[brazen schooner]

its says preprocess.log not found error

we can talk in dms if you could help. please

[ionic scaffold]

WTF?

fix the bot please

[wispy burrow]

Just talk here. So others could join to help.

[brazen schooner]

oh ok thanks

[rare goblet]

Ayo? @brazen schooner level 1 !!!

[ionic scaffold]

uh can I get help?

yall rude fr

[brazen schooner]

have you had success training models locally on mangio rvc?

[ionic scaffold]

don’t ignore me!

I need help

[brazen schooner]

im sorry, i cant help as clearly im also struggling with my own problem

[ionic scaffold]

ok but my problem is more important

wordband isn’t working

[green birch]

I just finished

[brazen schooner]

every problem is important tbh. i don't know what word band is. maybe someone else can help

[ionic scaffold]

your stupid it’s a bot in this server

[brazen schooner]

how?? im over here struggling

[ionic scaffold]

you knew that

[green birch]

What's the issue

[ionic scaffold]

uh I need help too?

[green birch]

Ir rude you don't need help u need manors and the block list

[brazen schooner]

im not usually on discord and not so familiar with everything on discord. call me stupid. you are such a rude person yourself. bold of you to assume im familiar with discord

[ionic scaffold]

I am not rude Wdym?

I literally just asked for help

sheesh

no need to overreact

@green birch how am I rude v

Huh?

how am I rude

explain

[molten pecan]

@half cove send the me when ai hub drama sticker

[half cove]

https://tenor.com/view/ai-hub-ai-kalomaze-kalomazing-drama-gif-17703842810846357578

[molten pecan]

Thank you 😭

[wispy burrow]

I have!

[brittle wing]

just look at this, why?

[brazen schooner]

bruh. we were nice. just because someone doesn't reach to your problem and doesn't understand something as I'm new here. doesn't mean you can call us names? ironic that we are over dramatic.

[green birch]

@wispy burrow ty for yesterday it worked sounds like garbage tho

[brazen schooner]

dont start a fight.

[molten pecan]

That's odd. It could be because the app and the files are two separate things. The files section is already built by default in huggingface, while the app is something the user makes. I hope that makes sense

[brazen schooner]

lucky

[molten pecan]

But try installing a dark mode extension for your browser

[ionic scaffold]

bro is a kid fr

I’m offended

lmaoo

bye losers and hope mommy takes the phone away from you.

🤣🤣

[wispy burrow]

Wordband is not "RVC".
You are in RVC Help channel...

[brittle wing]

This is not like the one I used the first time. when I came back, this happened

[solid shard]

yo guys, i saw that the google colab method doesn't work anymore, how can i make ai voices now?

[brittle wing]

okay

[molten pecan]

You could also try reducing the saturation from your monitor

[half cove]

We do not own wordband harish does @ionic scaffold

[molten pecan]

Or the contrast, I don't remember the name

[brazen schooner]

suddenly you sound like a kid bruh

[green birch]

It wasn't sudden

[molten pecan]

-rvc

[azure marsh]

Hey, @solid shard!

👇 Here are some useful resources to help you learn how to make ai covers

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

🎃 Spooky contest is active till 31st, check the announcement channel for further details.

[solid shard]

thanks

[brazen schooner]

this is all about rvc not wordband. but yeah go ahead be rude

[brittle wing]

i already try but ain't work

[wispy burrow]

Could you show us the cmd when do Process Data?

[prime sparrow]

there is like 20 sec of silence on 20 min sound file every 1 min is this gonna effect the model ?

[wispy burrow]

It will.

[molten pecan]

Have you restarted your browser? Have you turned on the extension?

[brazen schooner]

this right here is childish. and now im a child? im actually an adult. i would gladly help you but im sorry i cant. why do you have to be so mean? this is a positive friendly place. sorry we cant solve your wordband problem. jeez be nice

[wispy burrow]

Negatively.

[sand iris]

https://tenor.com/view/psp1g-what-gif-23713811

[brazen schooner]

im actually busy getting that for you all. sorry for the wait

[prime sparrow]

is it gonna be very bad ?

[wispy burrow]

You could use Audacity to remove all silence parts.

[prime sparrow]

ok thanks

[brittle wing]

yeah

[molten pecan]

My extension doesn't do that. I guess you gotta try with another one. But hey, it's a step in the right direction

Or change the settings of the extension

[brittle wing]

nah, Maybe it's because the app is having problems, maybe in the next update it will return to normal

[brazen schooner]

Here is the error I'm having

[rare goblet]

Ayo? @brazen schooner level 2 !!!

[green birch]

that loks like a browser problem

[wispy burrow]

There's an "effect" in the "special" category called "Truncate silence".
You'll want that.
It's quick and easy.

[molten pecan]

Gotcha. Good luck then. Download RVC locally if you keep getting issues

Or use Applio

[green birch]

you need to name the data sets

[brittle wing]

it's fine, if it still useable

[brazen schooner]

I did. I named it. All my datstes have a name. I'm just not sure if I should be using unzipped folder or zipped folder of my dataset

[proper shale]

zipped folder of dataset

[wispy burrow]

Bingo.
You have a space in 'RVC Mangio'
You can't have any spaces or any special characters in the path to the RVC folder.

[green birch]

i also found 1 is better than more than 1

[proper shale]

and that

[brazen schooner]

Oh. I feel stupid. I wonder if that would make a difference

[green birch]

100%

[proper shale]

Yeah. I trained 1,2K .ogg files once. Despite the 30 min dataset size, it took 1 min 30 secs per epoch

[green birch]

it does not allow that and dont run easy gpu / gui

[wispy burrow]

It certainly makes a difference.

[brazen schooner]

Lol. I'm sorry

[sand iris]

Which GPU ?

[proper shale]

T4

at batch_size 10

[green birch]

im taking off , but ya mind a dm ?

[green birch]

i just maxed everything but my rig is beefy

[wispy burrow]

I prefer not...

[brazen schooner]

One moment. I'll check

[green birch]

sigh ok boss

welp im stuck XD

[proper shale]

wait did u max out batch_size

[wispy burrow]

Well, this is a help channel.

[green birch]

i dont even know what my issue is to ask about its why i ask

[brazen schooner]

I feel this pain

[green birch]

i got the model to train 10,000 epochs but doesnt sound like the og the iron mouse or micky mouse or other tests i have downloaded sound perfect mine sounds fake hollow and wrong

but i dont know why

[brazen schooner]

I think it worked

[proper shale]

what was ur batchsize

@green birch

[green birch]

random is way smarter than me dude

[green birch]

max 50 or w/e

[wispy burrow]

10 000 EEpochs!?

[proper shale]

_ _

[green birch]

yup

[proper shale]

You should use like

[green birch]

only took a few hours

[proper shale]

8, 16 for batch_size

[molten pecan]

How do you even pronounce the word Epochs

[brazen schooner]

Oh gosh. I feel so dumb. It's so simple.

[green birch]

e- pok

[hard gust]

Is there a way to download old Applio version? since the last update it doesn't work for me anymore.

[wispy burrow]

That's waaaay too much. It is very likely to be overtrained.

[brazen schooner]

I'm new to trainning locally. It's my first time actually. Any tips guys?

İf I train to let's say 1000 epochs. How long would that take approximately?

[green birch]

guess i'll try less but i donno i think its the audio its why i ask for you ot listen sadly i cant tell the difference

[green birch]

took me about 20 min but again beefy

[brazen schooner]

Oh my datstes are around 30 minutes

[green birch]

i gotta jet back in a bit

[brazen schooner]

I used to train on colab. At a stage. I'd make rich models. Then yeah that issue happened lol

[brazen schooner]

I think I might be able to help

İf you want

[proper shale]

Depends on what your GPU is and your batch_size

[wispy burrow]

The most important thing is your Dataset.
Remove echoes/reverb.
And use remove as much silence as possible.
And if you have multiple short files make sure none of them are smaller than 1 second.

[brazen schooner]

Oh. Can this run in the background? And if so, must there be an Internet connection?

[brazen schooner]

Yeah i use uvr5 to clean my acapellas

İt works, it makes the vocals more clear. That's why the models become richer sounding.

[wispy burrow]

I just wished it has a batch inference...
I had like 400+ files...

[brazen schooner]

Wow

[rare goblet]

Ayo? @brazen schooner level 3 !!!

[brazen schooner]

You mean 400 vocal files in your dataset?

Or 400 datasets?

[wispy burrow]

They were all very short files too.
Longest one was just over 3 seconds or so...

[wispy burrow]

400 files in my dataset.

[brazen schooner]

Oh i had 500. And it was fine

I trained a model but this was on colab notebook rvc

I had 500 wav files in the dataset

[wispy burrow]

Did share your Model(s) on the old AI Hub?

[brittle wing]

Is it possible to merge two voice models that's the same voice but used different training datasets ?

[brazen schooner]

No I never share my models because then people steal them and claim them as their own. Not giving me credit

[brazen schooner]

İdk tbh. But it seems possible I guess. Maybe just train over the existing model. But this is an interesting question

[wispy burrow]

Yes you can.
In the 'CKPT Processing' tab there is 'Model Fusion' on local RVC.

[brazen schooner]

Well there you go. Lol thanks for helping him

Or her

[brittle wing]

would that merge the index aswell?

[wispy burrow]

I don't think it does...

[rare goblet]

Ayo? @wispy burrow level 19 !!!

[brittle wing]

I see .alright

[brazen schooner]

Off topic but I like your name and profile

Bread👍

[brazen schooner]

Did you?

[wispy burrow]

Only 1 so far. Sligs from the Oddworld games.
Not perfect but it is usable.
I am about to share my 2nd one.
Heart Aino from BlazBlue.

[brazen schooner]

Btw I have 12261 MB of Ram

Should I be alright with training?

[digital nacelle]

do you need windows to use this?

[brazen schooner]

I mostly make singers ai models. Those get used a lot more. But interesting

[brazen schooner]

Yes

[wispy burrow]

12 GB for your GPU?

[brazen schooner]

Yes

[digital nacelle]

the only dependency that requires windows is sox

[wispy burrow]

That should be good.
It's 2 times what I have...

[brazen schooner]

Oh it's actually my friends. My own laptop sucks tbh. It can't even do this

[brazen schooner]

What is sox?

[wispy burrow]

I think there was a Mangio for Mac.
https://docs.google.com/document/d/1KKKE7hoyGXMw-Lg0JWx16R8xz3OfxADjwEYJTqzDO1k/edit#heading=h.e9y3sfgsycb8

[wispy burrow]

Installing Mangio RVC Fork Locally (EASY METHOD)

[digital nacelle]

nm works great in linux

[rare goblet]

Ayo? @digital nacelle level 1 !!!

[brazen schooner]

where can one talk about UVR5?

is it okay to talk about it here?

[digital nacelle]

ultra violate radiation 5?

[brazen schooner]

ultimate vocal remover 5

it what we use to get clear vocals for training models

[low shard]

Yea,any problems with UVR ?

[copper pebble]

nick buon halloween

[low shard]

If its for a dataset its better to talk in #1159290066074607626 tho

[low shard]

Buon halloween anche a te rickie

[digital nacelle]

so it isolate's the voice only from a audio file?

[brazen schooner]

it does many other things, like extracting instrumental,drums,bass,other

[brazen schooner]

that too

[brazen schooner]

its about making and isolating stems from songs

[low shard]

We can talk abt it here

First did you see the guide for it?

[brazen schooner]

oh ok then. im trying to get the layers of vocals. you know the harmonies. ive seen people manage to filter out certain high notes and ad libs. i struggle to do it and dont know how

i only get the main vocals

[brazen schooner]

no actually

[low shard]

https://discord.com/channels/1159260121998827560/1167838554383208550 this can help you

[brazen schooner]

thanks ill look into it

[rare goblet]

Ayo? @brazen schooner level 4 !!!

[brazen schooner]

il also trying to extract certain instruments from instrumentals

i hope this can help. thanks so much

[rich yoke]

is there a faster way to add voice weights to Mangio than manually extracting them and putting the index and the other file in two seperate folders?

rvc gui seemed much easier with the 'import zip' button

and i feel like there's probably a button I'm just not seeing somwhere

[digital nacelle]

sox converts sound

[molten pecan]

In the huggingface spaces you just paste the huggingface/mega/google drive link of the model and that's it

[rich yoke]

idk what you're talkin about but thanks for trying to help, blud

[rare goblet]

Ayo? @rich yoke level 2 !!!

[digital nacelle]

does this only work on music

[rare goblet]

Ayo? @digital nacelle level 2 !!!

[proper shale]

nah, works with speech

[prime sparrow]

guys how can i change pth into index or zip file format ?

[proper shale]

you have to go back on the gui/colab u were in, insert same settings, don't preprocess or feature extract again, just load the dataset again (if ur on a colab) and then train the index

[prime sparrow]

thats a bit complicated is this the only way ?

[proper shale]

yeah

[prime sparrow]

im using colab what should i do i didnt under stand

sorrt if i bother u 🤡

[proper shale]

nawww its ok!

:)

[proper shale]

which colab r u using

[prime sparrow]

google

colab.research

[proper shale]

no but like

what colab notebook

[prime sparrow]

idk

:c

[proper shale]

is it RVC Disconnected?

[prime sparrow]

no

not applio

[proper shale]

u should be using that then

others just disconnect

and that one

despite the name

doesn't disconnect

#1159949278270193734

theres the link

[prime sparrow]

ok

[rare goblet]

Ayo? @prime sparrow level 2 !!!

[midnight geyser]

Hello, I made model, and there is like "robotic echo" in the background? Is this fixable by changing some setting, or did I trained it wrongly?

[digital nacelle]

can you do it by text or does it have to be voice to vioce

[midnight geyser]

I created it by voice to voice, but it can be text too.

[proper shale]

prolly a bad dataset

[proper shale]

has to be audio file

[rigid birch]

whats the best settings to use?

[proper shale]

in what exactly

and for what

[midnight geyser]

Okaay, i thought my dataset was good, the audio was clear.

[rare goblet]

Ayo? @midnight geyser level 2 !!!

[midnight geyser]

I will try again tmrw

[proper shale]

wait

how many epochs

[rigid birch]

im using the google collab version and just for music covers

[proper shale]

@midnight geyser

[midnight geyser]

250

[proper shale]

use rmvpe as your f0/pitch extractor

[proper shale]

what batch_size

[midnight geyser]

I think default (it was 8)

[rigid birch]

ok ty what about the other settings

[proper shale]

some of those u dont need to change unless ur result isnt that great

[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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[prime sparrow]

how can i get the index file again ? 🙂 its the disconnected one idk how can i get the index one

[digital nacelle]

is there a way to live transcribe

[midnight geyser]

I did 8, is it good or bad? 😅

[proper shale]

Not bad not bad, i think you should train more tho just to be sure

else add more clean data depending on size.

[mossy ridge]

can someone help me magio rvc

[midnight geyser]

how long the .wav should be? I saw it should be 10 sec files? What is the best lenght?

[prime sparrow]

guys how can i change pth to index im using disconnected collab

[proper shale]

They can be any length until like, 30 mins

[green birch]

i took 16min file

[mossy ridge]

can someone pls help with mangio rvc when i put my song with vocals. And refresh is never find with instrument mode

[prime sparrow]

what after training the index where can i find it on my drive 🙂

[wooden torrent]

Guys I’m having trouble with something

[midnight geyser]

So you dont have to cut the audio info smallet ones? 😶‍🌫️

[wooden torrent]

Hello?

[proper shale]

Nope

RVC does that automatically

[proper shale]

?

[midnight geyser]

Ohh, okay! Tysm 🫶

[wooden torrent]

I’m using Hugging face for some Ai Models and unfortunately the refresh option has been taking up an unnecessary amount of time than expected.

[proper shale]

Have you duplicated the space?

[prime sparrow]

can u help me i send a message above @proper shale 🙂

[proper shale]

oh

yeah

definitely duplicate the space

itll take a while to boot but youll be able to use it

[wooden torrent]

I tried to but it crashed and now it’s telling me that I already created this space.

When something crashed

[proper shale]

Just insert the same training variables, do the resuming steps, click on train index and click on export models

[mossy ridge]

can someone pls help with mangio rvc when i put my song with vocals. And refresh is never find with instrument versio

[prime sparrow]

it doesnt show that its exported its just say success

[proper shale]

check the folder then

in google drive

[prime sparrow]

there is alow of folders

[proper shale]

click on ur pfp, then on ur profile name and check the space

[proper shale]

In your Google Drive and not Colab

[prime sparrow]

it is the same wdym

[proper shale]

It should be inside logs

and in the folder with ur models name

[prime sparrow]

it ends with .index or diffrent thing ?

ok thx bro its on log @proper shale

[astral perch]

Rvc can also train a voice model from a dataset that speaks a different language right?

[proper shale]

Use the added index

not trained

[proper shale]

Yeah

[mossy ridge]

can someone pls help me mangio rvc idk how to fix it pls

[rare goblet]

Ayo? @mossy ridge level 2 !!!

[wooden torrent]

The damn page is saying that it crashed

[rare goblet]

Ayo? @wooden torrent level 2 !!!

[rare goblet]

Ayo? @tough ice level 1 !!!

[brittle wing]

I try to use a voice but it sounds always bad, I don't know how to use it...

[rapid lichen]

How to use the prebuilt models available on this discord

[wooden torrent]

I hate this website🤦🏽

[molten pecan]

Replay? What's that?

[astral perch]

Is there a software that can remove for example footsteps sfx and other sfx from audio because I tried several stuff but nothing has worked yet

[molten pecan]

If you want to get the best results, there are a few things to get into consideration. In what audio format are your vocals? In what quality? Is there any noise or rumble? Maybe harmonies? And is the model you are using a good one, with a clean dataset?

[molten pecan]

Yup, mvsep.com :3

I am not 100% what that separation type does though, but I guess it's self explanatory

Make sure to log in so you can select the output format and select wav. Wav format gives the best quality and the most precise results for making models or AI covers

[astral perch]

Omg thank you so much

I always used noise remover i think its called but that did nothing

Also is it true that mp3 format makes your voice worse

@molten pecan

[dull olive]

Can i use voice from serbian singer?

[molten pecan]

True. Each audio format does a different thing. mp3 and m4a compresses the audio quality, which means they lose quality and lose some data about the audio, in exchange of saving some space

[molten pecan]

@half cove can you give them permission?

[astral perch]

What do you use to record your audio from youtube. My google extention only supports saving in mp3 files

[half cove]

🤷‍♂️

[molten pecan]

https://y2down.cc/es/youtube-wav.html I use this one! Remember to select wav

But keep this in mind

YouTube already compresses the quality of their videos/audio once you upload something in it, so downloading in wav format won't do much if what you want is the raw, uncompressed video/audio

[astral perch]

I see

Thank you

I have one final question

[molten pecan]

You are welcome! ^-^

Ask away :3

[brittle wing]

Which is the best colab RVC for voice conversion (aside from ai cover generator)

[molten pecan]

-rvc

[azure marsh]

Hey, @brittle wing!

👇 Here are some useful resources to help you learn how to make ai covers

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[brittle wing]

-rvc colab

[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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[brittle wing]

Huh

[astral perch]

I have an audio of someone speaking but in the back ground there is voice singing but not but with words tho only like ha's and stuff. When I removed the music the vocals still stayed with the person speaking. How do I remove those vocals without remove the person speaking @molten pecan

I don't think those vocals are considered sfx

[dull olive]

Can somebody help me?

[molten pecan]

That's kinda tricky. I think you should use Ultimate Vocal Remover 5 and use the HP5 Karaoke or HP6 algorithm

You can download it locally on your PC or use it from the huggingface space, though the space is more unstable since it's on beta testing

It's free too by the way

[molten pecan]

What is it?

[dull olive]

Can i use voice from serbian singer?

[molten pecan]

Yeah sure

Search on #1159289700490694666 or on the website weights.gg

[dull olive]

That singer is not there

[astral perch]

@molten pecan does ultimate vocal remover have those same sfx remover models?

[molten pecan]

You can request for someone to do it on #1159289738314919936

[molten pecan]

It has like 50 different models 😭 so I'm sure you'll find the one you need

[astral perch]

thank you sm for your help

[molten pecan]

I think the hp5 or hp6 karaoke will do the job. One is for removing harmonies and the other for removing singing voices. So try and see which of the two work

[molten pecan]

You are welcome! You can also try googling which model does because I am not familiar with UVR5

Good luck!

[brittle wing]

maybe that helps clear out some questions

ah i cant send a screenshot here

[molten pecan]

Great then! Ping me if you need something else

[rare goblet]

Ayo? @molten pecan level 29 !!!

[molten pecan]

You can DM it to me if you want!

[brittle wing]

to me it doesnt let me for some reason

[molten pecan]

Are you below level 5?

That could be the reason

[brittle wing]

i sent dm with the screenshot ><

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

ah look, now i can post a screenshot xd

[dull olive]

Thank you!

[molten pecan]

What app is that?

[molten pecan]

Wait do you speak spanish?

[brittle wing]

si

[molten pecan]

Default - Altavoces

[exotic hornet]

is there a tutorial??

[molten pecan]

-rvc

[azure marsh]

Hey, @exotic hornet!

👇 Here are some useful resources to help you learn how to make ai covers

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[molten pecan]

Aahhh usás el cambiador de voz en tiempo real. Yo pensé que querías hacer covers de canciones

[brittle wing]

ahhh noo

[molten pecan]

No tengo ni idea el cómo usar ese programa, sorry 😭

Solo sé que en la parte de F0 Det., los mejores algoritmos son rmvpe y crepe

[brittle wing]

Sabes de alguien que me pueda ayudar con ello? :x

[exotic hornet]

-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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[molten pecan]

Me banearon por 30 días 😦

[wispy burrow]

That's W-Okada.

[molten pecan]

Si vas a hablar a una velocidad normal, usa el algoritmo rmvpe. Si vas a hablar de manera lenta, usa crepe. No es obligatorio pero te aviso eso si querés la mejor calidad

Así que si usás rmvpe por defecto 24/7 está perfecto igual también

[hardy smelt]

hey does anyone know why my AI replaced voice doesn't sync back up witht the original?

like its out of time

trying to line it up in software they get out of sync slowly and I dont know why, I dont remember it doing this before

[molten pecan]

Did you use the huggingface spaces?

Because that's a bug that happens sometimes

[hardy smelt]

I used the local Mangio-RVC-Fork

[rare goblet]

Ayo? @hardy smelt level 1 !!!

[molten pecan]

I see. That's odd then

Maybe you changed the tempo in the settings? I don't know

[hardy smelt]

I thought maybe I fucked with a setting but I dont think I did

[wispy burrow]

That's strange indeed.

[molten pecan]

I am not familiar with Mangio so I don't know how to help here. If it keeps happening use another audio format or try it in another device

[hardy smelt]

I'll play around and see if I can fix it

weird, I just restarted everything and it fixed it

[molten pecan]

Yup that's RVC 😭

[wispy burrow]

At least it's fixed.

[hardy smelt]

yeah 🙂

[viscid moss]

Yo!

Cual es el problema?

[brittle wing]

cuando lo intento usar mi voz suena muy mal... no sé cómo configurarlo

[viscid moss]

Activa Echo, Sup1, Sup2, cambia harvest por rmvpe, S Tresh dejalo en 0,0005, el CHUNK en 112 yel Extra en 4k

[astral perch]

Both speaking and singing in the dataset wont mess up the model right?

[molten pecan]

I don't think so, but I'm not sure

Ask on #📑│making-models

Or #1159290066074607626

[oblique venture]

I think singing is better, since it also included many tones and pitches, speaking mostly gonna be monotone so i dont think it gonna be great dataset

[brittle wing]

Ah? And upload images and samples acapellas?

[half cove]

#✨│announcements message

[wispy burrow]

An image is not an audio file...

[half cove]

Use external sources like YouTube

Also i think trusted members can post (i dont have model maker so I can't test)

[wispy burrow]

I don't want to rely on a link for the image.
If an image gets removed or replaced we lose the image/thumbnail...
I'd like to attach an image for the thumbnail.
I understand we no longer can attach audios though...

[half cove]

The thing is

On discord

Images and audio

Are both under files

So you can't just disable one without the other

A pain for real

[wispy burrow]

A pain indeed.

[proper shale]

Check #1159949278270193734

RVC-HF for example works for inference

[rare goblet]

Ayo? @brittle wing level 1 !!!

[grim bay]

Are there any new or fixed Mangio RVC Google Colab links?

[rare goblet]

Ayo? @grim bay level 5 !!!

[viscid moss]

try another colab for RVC

and duplicate the space in ur account

[grim bay]

Sometimes it will be 1000

[viscid moss]

https://discord.com/channels/1159260121998827560/1159949278270193734

[wooden torrent]

Can someone help me?

[violet glacier]

i have some questions about mangio,, i dont see my audio at the second "select audio section“? its always just the someguy or somegirl section

selection*

[grim bay]

Most of them just very slow or doesn't work

[viscid moss]

AICoverGen is the best one, but my version (spanish)

https://colab.research.google.com/drive/1qezaBvCn4eSG1U2L4a64tEHTynbIu8eN?usp=sharing

[potent root]

-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

[potent root]

i dont have rvc disconnected bookmarked

[polar elm]

File "subprocess.py", line 951, in init
File "subprocess.py", line 1420, in _execute_child
FileNotFoundError: [WinError 2] The system cannot find the file specified

What does this mean

[wispy nest]

does anyone know how to use the colab thing cause it aint working for me

like what do i need to download for it to work

[fickle mason]

was i born dumb or smth, the tuorial here " https://docs.aihub.wtf/vocal-cloning-guide/rvc-v2#how-to-install-easygui-locally" says to download this , "right-click on the RVC V2 file" BUT IT DOESNT EXIST???

it doesnt exist help

[rare goblet]

Ayo? @fickle mason level 4 !!!

[crude root]

Hi, does anybody know why my tensorboard is showing as triangles and not a line graph? I'm used to the Google Colab and this is really the first time I'm doing it locally

[crude jungle]

how do i make a RVC model?

[rare goblet]

Ayo? @crude jungle level 1 !!!

[crude jungle]

nvm i already saw a tutorial

[crude root]

Oh and also, no matter what I do the voice sounds like a static/ringing noise. I'd like some help if possible!

[potent root]

how do i make it such that colab doesnt flag me as inactive when im training

do i just click and scroll up and down or smth

how often do i do that

[brittle wing]

How can I convert the file to ONNX?

[strong belfry]

im assuming this is #🔍│help-w-okada. You have to upload the model to it first then click covert to onnx button

[molten pecan]

It's best if you don't use the Google Collab

Google has been banning the accs of people that uses it

So be careful

So you should install RVC locally or use the huggingface spaces

[strong belfry]

google colab (free) does not officially allow rvc.

It disconnecting you is basically them stopping you.

[molten pecan]

Yup exactly

[brittle wing]

where is this conversion button?

[strong belfry]

should be a button near the index sliders if i remember correctly. It wont show if its already a onnx model i think.

[brittle wing]

thank you so much

[potent root]

don't care cause i have like 20 disposable google accs

[molten pecan]

Ah

[bright veldt]

was the models archived???

[molten pecan]

Still, I think it's best if you use the other alternatives

Since I think they are more stable

[molten pecan]

Yup, check #1159289700490694666

Or the website weights.gg

[bright veldt]

all or most?

[molten pecan]

I think all

[bright veldt]

I see my models :DD!

[strong belfry]

I dont think it matters how many accounts you have if you cant get very far in training.

[bright veldt]

will i get my role back tho?

[rare goblet]

Ayo? @bright veldt level 1 !!!

[molten pecan]

Ask on #1159514067187277865

[viscid moss]

to recover your model-maker use /usermodels

in #1159290666019471421

[potent root]

ik what im doing lol i don't really need to get far into training since im only tryna do lightweight models

[bright veldt]

anyone got colab link?

i lost it a month ago D:

[potent root]

-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

[molten pecan]

-audio

[azure marsh]

Guides

📑 Perfecting Audio Isolation on Low-End Rigs: A Practical Guide (by Litsa The Dancer and Faze Masta)

https://docs.google.com/document/d/1HmkG9cmL8SLX7-vJcPT1-1KgUQtCrwXB8CicYmG4LW8/edit?usp=sharing

📑 Guide: Gathering and Isolating Audio (by SCRFilms)

https://docs.google.com/document/d/1wTJ_wutDqEtsA99BJOXDDGax25pPIDE84O5E2Rio5Qk/edit#heading=h.g896c5vbh0dv

📑 How to make a good model All-In-One guide by LUSBERT

https://rentry.org/RVC_making-models

▶️ Vocal mixing tutorial by Roomie

https://www.youtube.com/watch?v=rSxLF3kkBw0

Tools

Youtube Downloaders

• https://github.com/yt-dlp/yt-dlp
• https://github.com/dsymbol/yt-dlp-gui

Audio Separation/Isolation

• https://ultimatevocalremover.com/
• https://mvsep.com/
• 🆕 https://github.com/Captain-FLAM/KaraFan
• https://goyo.app/ (Note that this plugin is for clean ups only, like removing background noise, reverb, ambience or sustained sounds)')

Tip: Use -colab command if you're looking for audio separation/isolation colabs.

[molten pecan]

-help

[azure marsh]

Command has expired

[viscid moss]

-gpu

[azure marsh]

Ran out of GPU?

While training or inferring, if you encounter a dialog that tells you Cannot connect to GPU backend, your daily usage of GPU in Google Colab ended. You can either:

• Pay for more GPU
• Make an alt google acc

[viscid moss]

lol

[strong belfry]

Should the server really be telling people to abuse the system and make alts.

[grim bay]

I want something similar to this UI

[viscid moss]

That's not possible. Google has banned UI colabs

U can Hugging Face but.... it's too slow

[violet heron]

Google banned Gradio, which is was this ui uses

So you would have to do local or HF Spaces (HF Spaces only supports inference, which is making covers and stuff)

[bright veldt]

how do i use hugging face?

i only used Gradio so far

-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

[mild pollen]

i hate myself for being dumb but i cant really find anything on any help guides. ive done my audio, removed background with uvr, but how do i make this audio as a "dataset", convert in zip? shouldnt it be a pth file or something? any helpers

[glad zealot]

Not quite but tes

Yes

[tulip cloak]

DM me I am willing to make you guys model for free as long as it is interesting and easy for me. If you have the dataset, then definitely dm me otherwise even if you don't have a dataset you can still dm!

[tulip cloak]

Do you still have the dataset? I am willing to train

[glad zealot]

Kiichan Model OwO.. I'm currently making one but it's gonna be scuffed af so ye

[tulip cloak]

Do you have the dataset

[glad zealot]

On my laptop but it's dead rn sadly

[tulip cloak]

Ohk you can dm me whenever

[glad zealot]

I'm making on rn on colab using applio tho

[tulip cloak]

applio? Wasn't it banned a while ago

[glad zealot]

I have my own version of it on colab that works

[mild pollen]

i just wanna know how to do that by myself 😭 cant find anything how to

[glad zealot]

Applio devs doesn't update the colab version since google just banned it, so I made my own instead

[tulip cloak]

Wait what

[glad zealot]

Wut

[tulip cloak]

How are you even doing it

[glad zealot]

Magic :D

It have gui too :D

[upbeat ice]

ok so apparently whenever i copy and paste a model onto aicovergen it gives me "forbidden" error and idk why

anyone knows why that's been happening and what other alternatives to weights.gg?

[tulip cloak]

wait

how are you able to use it in mobile? it works on phone too?

if it does its just mind boggling

it will be so helpful for me

[glad zealot]

I'm using colab so everything I need like my dataset is on my drive

So it will work any pretty much anything as long as it has a browser

[tulip cloak]

ohk i am gonna try out rvcDisconnected on phone

since that's what i mostly use

[glad zealot]

Ye that would work

As far as I know the only thing that won't work on phone browser is the voice changer

[upbeat ice]

anyone?

[glad zealot]

Weights.gg seems to be working fine

[upbeat ice]

well when i copy and paste the model link it gives me "forbidden" 403 error now even tho the models were working and were downloadable before

[tulip cloak]

you can always check #🔍│find-models

[tulip cloak]

That is not on the weights.gg end it is probably because the model is deleted from the source

[glad zealot]

You can also go to huggingface.co for models

[tulip cloak]

thanks

[glad zealot]

Since as far as I know all models in weights.gg are in huggingface

[tulip cloak]

You mean like a repostiry? can you give me the link

[upbeat ice]

well i could've shown a picture of what's happening but image perms are off but it used to work and be copied and stuff onto aicovergen but it could be deleted maybe-

[glad zealot]

Direct link from weights.gg won't work

It need to be the zip link from huggingface

[glad zealot]

https://huggingface.co

[upbeat ice]

well it used to be there but i guess ima just try to download those and upload them on huggingface if that's the case

[tulip cloak]

So i just search a model there?

[glad zealot]

Yup

Don't reupload models made by other people, you can just upload the model directly to cover gen if you have it downloaded

[upbeat ice]

hmmm i guess so

but damn direct links from weights.gg used to be good tho, welp guess i gotta roundabout that

[tired rapids]

I've already ran everything, but why is this happening?

[blazing willow]

Hi
Can Anyone Help Me With Voice Model Of Akshay Kumar

[glad zealot]

I forgot the reason exactly but is your dataset a wav or mp3 file?

[tired rapids]

uh

[glad zealot]

Managed to get 250 epochs before it died :D I have auto save too so this is fine

[glad zealot]

Needs to be either mp3 or wav... MP4 does not work

[high night]

Wtf, on phone? That's crazy dedication

What do you even modify

I don't ever seem to understand

[glad zealot]

I was on laptop earlier but my sis had to use it so I switched to my phone

[glad zealot]

Mostly just links and names

[high night]

Oooh

That gives me some ideas

[tired rapids]

what's the average epochs for 5 recordings that are around 20 seconds each?

[glad zealot]

Not sure but try 300-500?

Or check the conversation guide on the pins on this channel

[tired rapids]

is that like the total of all recordings?

cause I have 5 recordings which are 20 seconds each

[glad zealot]

Ye

I forgot who said it but they said that using a 1 song dataset is better than separated

Not sure if it's true tho

[mild pollen]

what do i do with my finished audiofile? like how do i make this "dataset" a real dataset to work with? cant find anything about it anywhere

[rare goblet]

Ayo? @mild pollen level 1 !!!

[glad zealot]

Have you removed all the other voices, background music, static, reverb from your Audio?

[mild pollen]

yes did it with uvr

[glad zealot]

Dataset basically just means your clear audio, just the voice without anything else

[glad zealot]

Have you removed the silence?

[mild pollen]

yes, did that on audacity

[glad zealot]

Noice then you are done

What are you gonna use to train tho

Local?

[mild pollen]

but i feel like im missing some know how, like when i run The Mangio-RVC-Fork, i just dont know what to do with the audio file

yes local, trying to make a rvc model

[glad zealot]

On mangio it there should be a train tab right?

[rare goblet]

Ayo? @glad zealot level 30 !!!

[glad zealot]

Then somewhere in there it should be asking for the dataset

Not sure if it's the same as colab but is there a dataset folder on the mangio folder?

[mild pollen]

oh i have to put the audiofile in there?

[glad zealot]

Yup

Not sure if it will auto detect that it's there tho

On applio it needs a folder in the dataset folder called whatever your character is.. not sure if it needs that on mangio

[mild pollen]

well it still doesnt work, so i put the audio, i mean dataset, in the datasets folder and the path is showing there but it still gives an error

[glad zealot]

Is the path the audio or the datset folder

Iirc it needs to be the dataset folder not the audio

[mild pollen]

the path is on the datasets folder, and i put in the audio in, feels like this is wrong what i am doing

[glad zealot]

On the part that it's asking for the dataset path, you need to put the path to your dataset folder

I really shouldn't be the one answering this since I'm doing everything on colab

[mild pollen]

what colab are you using? i mean if it gives the same result maybe i should work with that

[rare goblet]

Ayo? @mild pollen level 2 !!!

[glad zealot]

I was using rvc disconnected befor

-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

[glad zealot]

@sand iris you know how to use mangio local? They need halp

[sand iris]

What's wrong

[mild pollen]

its fine i can try rvc disconnected

[glad zealot]

They have the dataset ready, just need to train it on mangio

[sand iris]

Are there spaces in any of the folder names leading to the dataset files ?

[mild pollen]

omg

[sand iris]

(also on the audio files themselves)

[mild pollen]

im so dumb yes, now it managed to spit something out

feels like i still have some errors or is this normal?

[glad zealot]

Nope that's not an error

You can continue

[mild pollen]

oh contiue on the bottom part?

[glad zealot]

To feature extraction now

[mild pollen]

this one right?

[glad zealot]

Yup

[mild pollen]

10h+ of reading and trying to find an error

all was the space of "New Folder"

....

[glad zealot]

Ahh

I think that could actually be an easy fix on the code but idk.. never opened it

[mild pollen]

ok i might be annoying but what should i do here now?

[glad zealot]

On the save frequency that's on every epoch you wanna save.. ex. Every 10 epochs so it saves epoch 10, 20, 30...

On the next one that's the total epoch you wanna train, howany epochs to be exact

I do suggest training the index first before the model tho

Not really any different but just to make it easier so you don't need to wait for it to finish to make the index

[mild pollen]

i dont have an index

[glad zealot]

It will make one

[mild pollen]

oh train feature index

[glad zealot]

The batch per gpu one is I have no idea :D I have mine on either 8 or 14 tho

[mild pollen]

so save frequency can stay at 5? and epochs? people do 200-300 right?

[glad zealot]

Ye thats like the safe spot most of the time

If you have the storage you can leave the save frequency at 5

[mild pollen]

index is already done, is it normal that it is that fast?

[glad zealot]

Yup

The model will take vereery long tho

So just wait till it finishes

[mild pollen]

god you two are lifesavers

[glad zealot]

Btw everything will be saved on a folder called logs on the mangio folder

There should also be a weights folder inside logs..that will be where your model will be saved

Inside the logs folder there should also be a folder called whatever your character was.. theren a file there that starts with "added" is the index file(accent file)

[mild pollen]

oh well i think i should have put in more gpu power, gpu is at 20%

but thank you for the information

thanks for help really

[glad zealot]

Aight imma go take a cat nap

[mild pollen]

take care

its finished i think, lol

[brittle wing]

nose porque no recibe bien lo que hablo

aun cambiando de configuracion

[gritty talon]

I've been trying to install EasyGUI, but I've been getting this error: ImportError: cannot import name 'I18nAuto' from 'i18n' (unknown location). I've seen 2 other people getting i18n-related errors, and 1 of them has this exact error. Is there any solution, or is there another GUI I should be using? I'm trying to run RVC v2 locally, I'm not open to switching to a colab.

[rare goblet]

Ayo? @gritty talon level 1 !!!

[sand iris]

Are you attempting to use easyGUI on mangio or the main RVC project?

easyGUI hasn't really been a thing for a while

[gritty talon]

I'm following the guide from docs.aihub.wtf

[daring verge]

colab downfall went wild

[thorn knoll]

hello, this error happened when I tried to run the ngrok cell block in hina's colab

cp: cannot stat '../client/demo/dist/temp/index.html': No such file or directory
Timer: 00:10:0941677
Timer: 00:10:10Traceback (most recent call last):
  File "/content/voice-changer/server/MMVCServerSIO.py", line 18, in <module>
    from voice_changer.VoiceChangerManager import VoiceChangerManager
  File "/content/voice-changer/server/voice_changer/VoiceChangerManager.py", line 9, in <module>
    from voice_changer.Local.ServerDevice import ServerDevice, ServerDeviceCallbacks
  File "/content/voice-changer/server/voice_changer/Local/ServerDevice.py", line 9, in <module>
    from voice_changer.Local.AudioDeviceList import checkSamplingRate, list_audio_device
  File "/content/voice-changer/server/voice_changer/Local/AudioDeviceList.py", line 1, in <module>
    import sounddevice as sd
  File "/usr/local/lib/python3.10/dist-packages/sounddevice.py", line 71, in <module>
    raise OSError('PortAudio library not found')
OSError: PortAudio library not found
Timer: 00:10:11

[hot violet]

do you want to send the modded applio gui colab? i want to try it

[brittle wing]

Hello, I'm using MMVCServerSIO, but I can't convert my voice, can you tell me how to fix it?

[tranquil cliff]

Hi

[brittle wing]

hi

[glad zealot]

Naah :D

[potent root]

-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

[astral jungle]

Yo

[copper pebble]

try to split the audio in 2/3 parts

what r u using??

@brittle wing

take ur time

ugh thats a lil bit old

https://huggingface.co/spaces/TheStinger/Ilaria_RVC

use this

should be better

[stable flint]

guys is easygui working for you?

[rare goblet]

Ayo? @stable flint level 1 !!!

[copper pebble]

its old

[copper pebble]

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=Yj-_npuA_HRB use this

https://discord.com/channels/1159260121998827560/1160926856548200531 heres the guide too

[stable flint]

im looking to only use inference. i dont know how to use this. is there a webgui version?

[copper pebble]

oh sorry

lemme get the links for ya

[stable flint]

thanks!

[copper pebble]

https://huggingface.co/spaces/TheStinger/Ilaria_RVC you can use this

really easy

https://discord.com/channels/1159260121998827560/1167844377046040769 heres a guide

[stable flint]

thanks. converting is taking a while. is this normal with this version?

i also got an error

[copper pebble]

-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

[stable flint]

yeah its not working.

[copper pebble]

if you prefer colab use the first link of this embled .. for ilaria rvc if you got errors or the convert process its too long split the audio in 2/3 parts

[copper pebble]

wt?

[stable flint]

errors. and my file is pretty small already, 41 seconds

im looking to convert audio files, so the aicovergen is not needed for my use

[copper pebble]

ilaria rvc?

[stable flint]

yup

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=F6zbsuNs6xZF i used to use this but it stopped working today

[copper pebble]

strange

mhhh

idk it should work

[rare goblet]

Ayo? @brittle wing level 2 !!!

[copper pebble]

did u follow the guide??

its strange cuz i just did a cover with that

@brittle wing how much was the audio long??

mhh

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

if the problem persists split the audio

in 2 or 3 parts

it should work

[copper pebble]

does it persists ??

[stable flint]

it worked now

it took a bit

had to refresh a couple times

[copper pebble]

nice

[glossy coyote]

the huggingface version is slower probably because it's using cpu(free) instead of gpu

[copper pebble]

lemme know if u need help

yh i think

[stable flint]

it actually messed up the conversion tho. it made the audio faster than it should

[copper pebble]

mhh

[stable flint]

an example

[copper pebble]

whats going on today with ilaria rvc

wth

its weird cuz i did a cover like 10 minutes ago and it turned out perfect

[stable flint]

yeah today somethings wrong. easygui stopped working too. so idk whats going on

[rare goblet]

Ayo? @stable flint level 2 !!!

[stable flint]

maybe colab has issues today?

google colab in general

[copper pebble]

well easy gui its old

[stable flint]

it was perfectly fine until today tho

[copper pebble]

yh but you could be banned

i mean your google acc

if u stay to longer in it

[stable flint]

im paid

[copper pebble]

ooooohh

[stable flint]

yeah

[copper pebble]

then its good

[wise karma]

how do i run rvc-2 locally with nvidia, i followed the guide here but it doesn't work
https://docs.aihub.wtf/vocal-cloning-guide/rvc-v2
when i have selected the inferencing voice, some of the ui change to displayed "queue: 1/1 | XXXs" with XXX go up to hundreds seconds

-local

[azure marsh]

All Working Local Forks 🖥️

• Applio-RVC-Fork, 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 Huggingface

Need some help? 🤔

You can find more info on the "❔┋guides" channel. If you can't find your answer, feel free to ask for help in "help-rvc". Credits to Faze Masta and Antasma for compiling these links.

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

[glossy coyote]

i think it's saying your model did not download correctly

[copper pebble]

da quel mess penso tu sia ita

hai risolto o ancora persiste?

[wise karma]

-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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[rare goblet]

Ayo? @wise karma level 1 !!!

[restive marten]

How to get model URL if this model is hosted on huggingface only

[proper shale]

Search for their username (or the model's name) and right click on the download button to copy the download link

(once u reach their model page)

[restive marten]

I'm on android

[keen mica]

whats the ideal dataset length before it gets useless to add any more

[rare goblet]

Ayo? @keen mica level 1 !!!

[marsh grove]

Yo yo yo it's ya boy Shrimpai in the house

[proper shale]

Just tap and hold on the download button

[proper shale]

After like 2 hours RVC won't use all the dataset

[lost lantern]

yo this exists?

thanks

[fleet raven]

Hi! When I hit start training my model the cmd says max value is tensor (and some random number). Can anyone help?

[keen mica]

Whats the ideal length i meant before the different in quality is not noticeable

[proper shale]

If you have consistently good stuff? Idk, probably like 40 mons

mins*

[keen mica]

so the difference in 15 min dataset vs 40 min is noticeable?

[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

🇫🇷 France

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

🇩🇪 Germany

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

🇮🇹 Italy

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

🇯🇵 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

[proper shale]

It depends on the type of voice

[restive marten]

thanks i already learned it but thanks anyway

[rare goblet]

Ayo? @restive marten level 1 !!!

[proper shale]

Glad to have helped :)

[stable flint]

any update on a colab link for inference that works?

[glad zealot]

cover you can either use the no ui ver or the ui ver

-cover

welp that ded

[stable flint]

?

[glad zealot]

https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb

that one for no ui

[stable flint]

i want to convert audio files

[glad zealot]

wdym by convert?

make ai sing song?

[stable flint]

no just swap a voice to another voice

[glad zealot]

yes that is a cover

bruh

[glad zealot]

use this one for no ui

https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb this for ui

basically the same thing anyways

[stable flint]

should i keep pitch change 12? i dont want to change the pitch

[glad zealot]

12 gives you more control to the pitch so i guess so

[fickle mason]

The conda Installer for RVC studio isnt doing anything but showing me a empty cmd HELP?????????????

herro?

sad day

[stable flint]

not working

[royal valve]

I use first time applio-rvc-fork(amd support) and something got wrong. when i tried to train model applio web.bat show up these codes. What's meaning and how to solve this problem? Traceback (most recent call last):
File "D:\ApplioRVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 18, in <module>
hps = utils.get_hparams()
File "D:\ApplioRVC\Applio-RVC-Fork\lib\infer\infer_libs\train\utils.py", line 412, in get_hparams
hparams.collapse_threshold = args.collapse_threshold * 0.01
TypeError: unsupported operand type(s) for *: 'NoneType' and 'float'

[rare goblet]

Ayo? @royal valve level 1 !!!

[still crystal]

why does the model training stop at 1 epoch?

[dense crane]

Don't use Applio

[dense crane]

What do you mean by that

Send a SS of you console

[still crystal]

ok wait

[sand iris]

Ya can't train with AMD unfortunately

[dense crane]

How the hell did you know they were using an AMD card

[sand iris]

(amd support)

I'm guessing

[dense crane]

OH

I thought that said (and support) I'm stupid

[tribal ledge]

hey guys, just one question, where do I go to clone a voice? Before I went to the easyGUI collab but now it's completely different, I have no idea how to use it

[still crystal]

this?

[copper pebble]

-colab

[azure marsh]

Suggestions for @tribal ledge

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

[copper pebble]

use rvc disconnected

https://discord.com/channels/1159260121998827560/1160926856548200531 heres a guide

[tribal ledge]

thx bro

[rare goblet]

Ayo? @tribal ledge level 1 !!!

[still crystal]

@dense crane

[sand iris]

Aye... training on 4GBs GPU won't work

[still crystal]

Oh ok

[sand iris]

Though you could always (ab)use colab

[still crystal]

How do I do that?

[sand iris]

-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

[sand iris]

"RVC Disconnected"

[digital nacelle]

where's uvr5 model

[sand iris]

The UI is nowhere near as nice, but you'll be able to train models at least

[still crystal]

Is there a guide?

Link

[tribal ledge]

guys, what is the difference between harvest and rmvpe? does someone have better quality than the other or is they faster?

[sand iris]

https://discord.com/channels/1159260121998827560/1160926856548200531

[still crystal]

Ok thanks

[dense crane]

https://tenor.com/view/omw-to-spread-misinformation-all-over-the-internet-gif-26525511

I use to train on a 4gb 1650 Ti laptop edition

[sand iris]

How slow is it

[stable flint]

why did old colabs did and inferences on webgui arent working?

[sand iris]

Has to go within shared memory which is no good

[dense crane]

Not that slow as long as you keep your datasets under 5 minutes

[copper pebble]

rmpve is better

[glad zealot]

from what ive heard rmvpe is faster and clearer than harvest

[inner hatch]

Hello, Any advice on what i should do to start making models locally? Like should i use RVC, Colab, or Applio?

[molten pecan]

RVC and Applio are the best ones I think

I recommend RVC since its much easier to install

[scarlet pumice]

are there any working colabs / spaces where you dont need to extract the accapella

[stable flint]

Hi guys I’ve tried many links but inference isn’t working correctly. Can anyone convert audio files for me for the time being? Will pay if anything

I used to use easygui but it’s not working today idk why

[molten pecan]

https://huggingface.co/spaces/r3gm/Aesthetic_RVC_Inference_HF

I use that one

[thin stump]

Are you training voices? Harvest is the worst and Mangio crepe or Crepe is the best quality

[stable flint]

Haven’t tried this one yet I’ll get back to you

[molten pecan]

There are lots of ways to use the inference and some of them are unstable, so you gotta find which one works for the time being

[stable flint]

Do u know if old collabs have finally died?

[molten pecan]

It's really easy to use, no need to pay someone for it, don't let anyone take your money

[stable flint]

Cause I used easygui rejects for a while and it died today

[molten pecan]

Yup, Google has been banning the collab spaces

[stable flint]

Even if you pay?

Cause I paid for collab pro lol

[molten pecan]

Uhm I think so?

[stable flint]

Damn is it a legal issue?

[molten pecan]

Yeah I guess so

[stable flint]

Insane

[molten pecan]

Why the friend request though

[stable flint]

In case u want a job lmao

[rare goblet]

Ayo? @stable flint level 3 !!!

[molten pecan]

Ah 😭

[stable flint]

I sometimes need ppl for help with this

[molten pecan]

I see

[sand iris]

no

Also when have they been banning them last

If it was like two months back we have it figured out

[molten pecan]

Shay, trust me, this is much easier than it seems, right now the technology for AI covers is primitive and unstable, but it's really easy to use if you find the right way. Don't waste your money on people if you don't have to

[molten pecan]

I heard a lot of people say so

[sand iris]

They are banning anything using external web pages as UI, for free tiers.
Paid users can use whatever they want

[molten pecan]

I never used collab personally so I wasn't sure

[sand iris]

Notebooks that don't rely on external pages are fine and allowed

[molten pecan]

I have no idea what you are talking about 😭

[stable flint]

How about local. I’m a Mac user and I know before it was pretty difficult to run local on Mac. Has this changed? What would you recommend I use?

[molten pecan]

Why would you buy a notebook lol just use .txt files

[stable flint]

I don’t really need to train models I just want to convert audio files / use inference

Download models

[sand iris]

No, you are not getting me.

[sand iris]

Not sure how stable things are nowadays to be frank, not many have tried lol. But it should be supported

[molten pecan]

I think the huggingface spaces are compatible with Mac. But if you want to run it locally I think Applio is compatible with Mac

[ruby narwhal]

why my RVC: Res always increase

anyone please help T T

[molten pecan]

Res always increase?

[ruby narwhal]

yes

[molten pecan]

Wdym by that?

Gdr what are you cooking?

AI Hub's declaration of independence?

[sand iris]

Google colab (the platform), has a free tier available to everyone, and also paid options for those wanting unlimited access or better hardware.

• Free users aren't allowed to abuse the platform by running code that uses an external web page as user interface. Which is why people have started writing codes that don't need external pages, so they're allowed for free users

• Paid users do whatever the fuck they want

[molten pecan]

Thank you gdr

[sand iris]

Should be heading to #🔍│help-w-okada (unless you aren't running w-okada/voice-changer)

[molten pecan]

But why would someone pay for paid collab when applio is free?

[still crystal]

do u know what im supposed to do here?

[sand iris]

Because they don't have the hardware to run it on their own machine

[molten pecan]

What about the huggingface spaces?

[sand iris]

So there's supposed to be settings that you have to modify

[molten pecan]

I can use them just fine on my crappy phone

[sand iris]

Cant train models on that

[molten pecan]

Ah

That makes sense

What's more expensive? Paying for paid collab or for a commission?

[sand iris]

¯_(ツ)_/¯

[still crystal]

these?

[sand iris]

Ya

Been a while since I used that, but you should have a .zip file ready in your google drive where it expects it

[still crystal]

do i need to write anything here or no?

[rare goblet]

Ayo? @still crystal level 2 !!!

[sand iris]

the name of .zip file

[still crystal]

ok

[inner hatch]

Yeah, i was looking at the RVC Model train guide on aihub.wtf and it said something along the lines of using appilo cuz it was better

[royal jacinth]

is there a way to train models using amd gpu because I heard there is

[untold zodiac]

hello I'm new here can anyone give me a step-by-step tutorial of how to make my own voice model and convert someone else voice into mine

[copper pebble]

you need to do a model of your voice

1st of all you need to record min 5 of yourself: talk or reading or singing

[molten pecan]

-rvc

[copper pebble]

wth?

[copper pebble]

this will be your dataset

you can use rvc disconnected to train it

-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

[copper pebble]

https://discord.com/channels/1159260121998827560/1160926856548200531

heres a guide @untold zodiac

dm me for any problems

[untold zodiac]

ok thanks for the help

[copper pebble]

[tranquil cliff]

rickie volevo infoemarti che se fai''/guide'' esce anche la mia guida

cosi è molto piu semplice farla vedere alle persone

[azure marsh]

RVC Guides (How to Make AI Cover)

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: Google Docs

Create a model with RVC disconnected (colab)

Link: Google Docs

Credits: Nick088, Angetyde and Kanav

[copper pebble]

ah okok

grz bro

[tranquil cliff]

prego

[brittle wing]

colab

[proper shale]

colab?

[azure marsh]

RVC Guides (How to Make AI Cover)

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: Google Docs

Create a model with RVC disconnected (colab)

Link: Google Docs

Credits: Nick088, Angetyde and Kanav

[uneven elk]

I never used Google colab before @proper shale

[proper shale]

It should be simple tho

[uneven elk]

But there's so much codes

[proper shale]

Just click on the play buttons to run the step, wait for a green checkmark

[uneven elk]

Oh k

[proper shale]

Until the last step where u gotta click on the link in the output

Thats where the AICG GUI will run

[uneven elk]

K

@proper shale

[proper shale]

nooo wait thats the wrong one

thats for training

[uneven elk]

Oh?

[proper shale]

https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb

this is the one

[uneven elk]

Ty

It's still loading but it says This share link expires 27 hours. for free permanent hosting and GPU upgrades

[proper shale]

Click the link

Public link

[uneven elk]

I did

It just takes me to Spaces

Like what next

[rare goblet]

Ayo? @uneven elk level 2 !!!

[uneven elk]

You're there? @proper shale

Idk what to do next?

[proper shale]

Now you just do the thin

It should work

[uneven elk]

When I clicked the public link, it showed me this?

[proper shale]

huh

that

wait have u clicked on the right one, ends with gradio.live

[uneven elk]

Ohh

Nvm! XD

[proper shale]

Happens lol

:)

[uneven elk]

XD

And you're sure it will work?

Because I really don't want errors

[proper shale]

Yeah

Unless u do smth wrong

[uneven elk]

K

Alr then

Ima generate

[proper shale]

It'll do it in 150-200 seconds

[uneven elk]

Oki

So not long then