✨│ai-help

[blazing garnet]

k then it should work try it

[brittle wing]

I'll see about that, let's see

[blazing garnet]

this too advanced for me

gotta call reinforcement

@proper shale wake up

[fierce crown]

Hey i was wondering if there is a way to get less delay. Currently it takes about 2 seconds after i say something until i hear it

[proper shale]

did you preprocess it?

[proper shale]

what gpu do you have?

[fierce crown]

4060 (or 4060 laptop version)

[blazing garnet]

doesnt one click training not req preprocessing?

[blazing garnet]

u rich

[proper shale]

oh? hmmm i suggest reducing the chunks if they're too high

[fierce crown]

we found a demo model that was half price lol

[proper shale]

there isn't a one click training option on applio thk

[blazing garnet]

bruh

[blazing garnet]

oh...

i need to learn more

@brittle wing did it work?

[brittle wing]

Doing the process and training, i'll tell once it's done

[blazing garnet]

k

[blazing garnet]

was ur problem solved?

[fierce crown]

Im still testing the diffrent chunks to see which one is best

[blazing garnet]

i am so bored

[solar jacinth]

wait , i will try asking commands to chatgpt

[rare goblet]

Ayo? @solar jacinth level 11 !!!

[brittle wing]

And also, is there a way to change applio's language? so I can put in english?

[solar jacinth]

192

[proper shale]

i think u can

[brittle wing]

I wish, because the portugese translation is so confusing and I was used to the english language in rvc lol

[fierce crown]

It got over double as fast thank you @proper shale

[proper shale]

you're welcome :)

[sand iris]

Don't think it'll work chief

unless you are really precise about the paths, it cant guess them for you

[golden dove]

I'd like advices to make a quality dataset. I'm a beginner and trained a test model on a ~40min dataset.
I'm processing audio to get more data of better quality, but there are a lot of things I don't know:
(for context, I'm extracting vocals from a female japanese singer in order to make covers)

• should I remove breaths ?
• should I remove lower db parts ?
• how long should the dataset be for best results?
• how varied should the dataset be (calm / fast / high pitch) and should it be proportionally of equal length ?
• Should I remove parts of a song that repeat the same lyrics ?

[proper shale]

1. no
2. yeah, keep everything normalized
3. it really depends on quality, normally 10 mins is fine for most models
4. try to put lots of variations of pitch mainly, and it doesn't need to be proportional
5. yes

[golden dove]

Thanks !
Shouldn't longer datasets produce best results, even with very low difference ?

[solar jacinth]

ohh also , how many epoches should be trained for -40mins dataset

[proven hill]

HAPPY NEW YEAR AI HUB!! 🥳

Buy e-girl, mommy and v-tubers voice models here: https://ko-fi.com/ilariaowo/shop

You can also find FREE voice models!

If you DM me and pay with PayPal you will get a 20% discount! So do not esistate!

[solar jacinth]

yayy it works , import shutil
src_path = r"/kaggle/input/model-pth/Gigi_e300_s75900.pth"
dst_path = r"/kaggle/working/HTweaks/weights"
#shutil.copy(src_path, dst_path)
#print('Copied')

shutil.move(src_path, dst_path) # Use shutil.move() for moving
print('Moved')

[golden dove]

I'm also curious about that as I don't know how to test lower epoch models (D/G files not accepted in rvc)

[solar jacinth]

ohh

[glad zealot]

you could just have use mv

[solar jacinth]

ToT , my copy path code is longer than your real working codes

[solar jacinth]

I will try

[glad zealot]

i would just do !mv <file> <directory to place file> or if that doesnt work use !cp -f <...

[trail charm]

hello the ai model im using doesnt sound like artist at all , what could be the reason?

[solar jacinth]

like this one ?

!cp -f /kaggle/input/my-recordings/ai slavery.m4a /kaggle/working/HTweaks/datasets

[solar jacinth]

which one ?

[trail charm]

im using xxxtentacion rap + sing and only sing voice model

but there is not a single thing sounding like xxx , yesterday i didnt had this problem

[solar jacinth]

!mv /kaggle/input/model-pth/Gigi_e300_s75900.pth /kaggle/working/HTweaks/weights mv: cannot remove '/kaggle/input/model-pth/Gigi_e300_s75900.pth': Read-only file system

[golden dove]

I used to clean a full song to add to my dataset.
From what you told me, it seems like picking 5-10 seconds clips of the best quality from each song would be better.
Could you confirm ?

[rare goblet]

Ayo? @golden dove level 1 !!!

[pure scroll]

i guess the voice model wasn't able to understand most of the voice patterns

[golden dove]

The model was very meh, didn't use the full range the voice of the artist can, and a bit robotic at some points.
The issue is that I'm not sure about how much to train it, so I guess I'm trying to get a perfect dataset so I can be sure errors would come from training parameters

[trail charm]

yo this is drivinng me crazy, every ai model i use doesnt sound like the artist , im using text to voice to first try the model before using full audio but the text already sounds dogcrap. Yesterday i didnt had this problem could it be that the RVC is bugged out or something? (im a complete newbie to all this stuff)

[rare umbra]

Could someone help me solve this?

[golden dove]

I can only hazard a guess, did you run install_Applio.bat and did it throw any error ?

torch doesn't seem to be present, the script is supposed to install it

[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

[agile storm]

guys hows that website called that had a lot of mdx and uvr models to separate instrumental from vocals?

[brittle wing]

Anyone know how to fix this since I haven't received my stuff from yesterday

[violet heron]

Try again

[brittle wing]

K

[rare goblet]

Ayo? @brittle wing level 3 !!!

[violet heron]

Wordband has that issue every now and then

[brittle wing]

Nope

[autumn wagon]

Hi, I'm using the cover website there, I've already chosen a voice, and I want to transform it into Ariana's voice, but when I go to select the file it doesn't give me the option to click, I've already saved it in video and audio too, but none of them work, how do I do?

[violet heron]

Drag the files from the files app into the box

[autumn wagon]

I don't know how to do this, help

very confusing

drag like this?

?

[violet heron]

On the files app

Not on the upload

[autumn wagon]

OOOMGG

thank you so much!!!!!!

[light pagoda]

the dataset file is gone?

Does anyone know how to fix this

[proper shale]

No? Try to use as much as you can from songs, especially if they have variations

[earnest musk]

Hmm... doesn't look like you have a Weights account. Sign up or link your Discord account below.

Hmm... doesn't look like you have a Weights account. Sign up or link your Discord account below.

Hmm, no models found for that name. 😔

Hmm... doesn't look like that's an audio file. Please try again.

🎉 Your Creation was started! It's currently queued at position 37.

Weights AI Cover | Max-sings-_Hardware_-from-Schoolhouse-Rock..mp3

By: Deez Nuts

[cyan spear]

https://cdn.discordapp.com/attachments/1159290139609137264/1191143122089025586/image.png?ex=65a45d7a&is=6591e87a&hm=9d798ff579c78a3ca5a43905ba773a426e3bb7d8b78f6aebfec19b5e9720ae20&

still having this issue

any fixes??]

[hidden drum]

llaria rvc is a thing?

[jade marsh]

when installing fairseq with "pip3 install fairseq" (using python 3.9.0)

[molten pecan]

https://rentry.org/ilarvc_inf_guide

My guide ;3

[molten pecan]

Which website specifically?

[autumn wagon]

I already got it, Ty ❤️

[molten pecan]

Yeah but, which website did you use specifically? Some websites are a little crappy and I know the best alternative >_<

[sand olive]

when i open the start.htp.bat nothing happens... please 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

[mystic tangle]

Best settings for a 3060 rtx?

nvidia

[hidden drum]

Thanks

[molten pecan]

yw! Have fun

[kindred mural]

what's the difference between w-okada and rvc?

[rare goblet]

Ayo? @kindred mural level 2 !!!

[brittle wing]

Aaaaaaa plz can someone make the ai sing it I'll send them the thing to make the ai sing

I just need 2 ais to sing the song

model:bora_karaca_cojum_diptally_hall youtube:https://youtu.be/rfUeWe7u364?si=TeOguhvXrDwGumfR

model:joe_hawley_tally_hall youtube:https://youtu.be/rfUeWe7u364?si=1CgdA38yIbpwBULA

Can someone do it for me and send me the ai vocals in both and the instrumental

@violet heron can u do it?

[molten pecan]

RVC is an advanced voice cloning AI software. You use RVC for training (making models) or inferencing (converting vocals through audio files, like AI covers).

W-Okada is a software for converting your voice in real-time (not through audio files) as you speak through the mic. W-Okada is built with RVC among with other apps

[kindred mural]

Ah I see, thank you

[molten pecan]

yw!

[brittle wing]

K but what channel would it be for broke ais

[brittle wing]

Like this one

Bc I just need the help now

To create the cover

[molten pecan]

https://rentry.org/ilarvc_inf_guide this one works in any device

[brittle wing]

Ok... But I need these pasific ais

If u are listening u would understand what I need

[molten pecan]

Pasific ais?

Wdym?

[brittle wing]

Scroll up

[brittle wing]

.

[brittle wing]

.

These 2

[molten pecan]

Why can't you do it?

Are you getting an error?

I can help you out on that

[brittle wing]

.

I haven't generated anything all day and haven't received my covers

[molten pecan]

Why can't you do the cover yourself?

[brittle wing]

Can u read

The photo says why

[molten pecan]

There are plenty of more tools for making AI covers other than Wordband

[pearl loom]

Hey Julia, after duplicating, is there an easy way to reaccess that after closing the tab out?

[molten pecan]

Yeah! I mentioned it in the troubleshooting section

Scroll down :3

[pearl loom]

Okay, I'll double check there. Tbanks!

[brittle wing]

Yes but I need the ais especially the bora one I can't get these specific ais anywhere else

[molten pecan]

Wdym by ais?

What's that?

[brittle wing]

Ai

With a s at the end

[molten pecan]

You can't get the link for the AI voice model?

Did you search on weights.gg yet?

[brittle wing]

What's that also I've sent the ais if scroll up and read u will see them rlly easily

[brittle wing]

.

Julia ik ur a mod and helper but ur not rlly being helpful rn plz just help me

It's not that hard to understand

[molten pecan]

I am not understanding what you are trying to say 😭 also I'm not a mod

You couldn't find like for the Bora AI voice model?

[brittle wing]

I need u to use the ai models and make wordband to create the covers bc I cant

[brittle wing]

This the only one I have

[molten pecan]

You know you don't have to use Wordband for making AI covers, right?

[brittle wing]

It's a rate finding

[molten pecan]

Did you search the model on weights.gg yet?

And #1175430844685484042 ?

And applio.org/models?

[brittle wing]

What is weights. GG

And voice models wouldn't have tally Hall ones

These are rare ais can't u just make the ai generate them it's not that hard I can't do it bc the ai won't let me

I just need both vocals and the isntrumental

[molten pecan]

A website where almost all the models are stored

You can make any voice model yourself too

On any device

You can also extract vocals and instrumental on any device

No need to use wordband

[brittle wing]

Yes but I don't want to make ais I just want to use the ais that already are made

[rare goblet]

Ayo? @brittle wing level 4 !!!

[brittle wing]

Plz just listen it's easy to understand it's not rocker science

Just make the ais cover the song then send the instrumental and both vocals to me

[molten pecan]

Wordband doesn't allow you to make covers when you are not logged in to Wordband

Are you logged in already?

[brittle wing]

Yee

Yes I am

[molten pecan]

Alright

[brittle wing]

Your read the thing ur making it harder for ur self

[molten pecan]

I'll see what I can do

[brittle wing]

If I had the ais on a website that just has a copy link here thrn generate button I wouldn't be going through all this

[molten pecan]

https://www.weights.gg/models/cln8h156b01otwsroksidn2o2 found your voice model

[brittle wing]

That's not what I need tho

I don't want that

[molten pecan]

You can make unlimited AI covers with Ilaria RVC

[fleet cedar]

is that e girl

?

[molten pecan]

No need to use wordband

[brittle wing]

Yes but I just need it done now

Plz juilia ur making it hard for useful it's a simple task to do my dog could probably do it if it had hands

Plzzzzzzz understand me

[molten pecan]

Helpers don't exist to take requests for AI covers, you know that, right?

[finite galleon]

Explain what u want

[molten pecan]

An AI cover. He wants to exclusively make it with Wordband because it's urgent apparently

[brittle wing]

Basically I need 2 ais to sing a song and I just need both vocals and the instrumental

[brittle wing]

.

Read above dot

[finite galleon]

Oh you dont want it combined?

Like seperate files?

[half cove]

Just split it using uvr or whatever

[brittle wing]

Just seperate audios

[half cove]

Mvsep for online

[finite galleon]

.

[half cove]

Use mdx23c

[brittle wing]

I don't want to do that bc I don't need to go through all that just download the vocals and instrumental

[pearl loom]

@molten pecan thanks again for pointing me in the right direction, you were very helpful

[finite galleon]

Huhhh?

Wdymm

[brittle wing]

Why is everyone here just making it harder for themselves it's a simple task to do I would do it but wordband is broken

[half cove]

What's broken about it?

[molten pecan]

You are welcome!!

[brittle wing]

scroll up

[finite galleon]

Your making it harder for your self if you dont want to do this and that

[brittle wing]

Basically I need the ai to use make both models sing the same song then I need the acpellas from Both models and the songs instrumental

[half cove]

You have used your limit I think it resets in a hour or day

No actually that means you already have one in queue

[brittle wing]

That's not what happened

[half cove]

I think its just taking a long time

[brittle wing]

Bc I have generated it yesterday and still haven't received them

[half cove]

You can try uhh

[brittle wing]

I've tried

[half cove]

Aicovergen

[finite galleon]

Try astralabs

[half cove]

Or @harsh scarab

[finite galleon]

Its the same as wordband i think

[brittle wing]

Can't u just do what I need and stop asking me to use different websites

[finite galleon]

Nope 👍

[brittle wing]

Why it's not that hard I would rlly do it myself but I can't because of the broken robot in this server

[half cove]

If the bots broken it ain't the servers fault

We don't make wbb

[brittle wing]

I mean it kinda is the severs fault for using it and not using a different one that has unlimited useage

Also the task in asking isn't that hard to do

[finite galleon]

[half cove]

We do have @harsh scarab

And huggingface

And colab

[finite galleon]

How are you still sane

[half cove]

I'm too far gone

[brittle wing]

How do u still have a brain cell left

It's a easy task to do

[finite galleon]

Then you do it dumbass

[tired kraken]

If it’s such an easy task, why not do it yourself?

[brittle wing]

I can't do it dumbfuck the ai won't let me

[brittle wing]

.

[finite galleon]

Its not only wordband bot that generates this shit ffs

[brittle wing]

Read the ais message

[finite galleon]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• 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

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

[finite galleon]

Click the aicovergen link

Its simple as fuckin pressin buttons that my water can do it

[brittle wing]

Ye but I need the Spasific models

[brittle wing]

Water can't touch electricity

[finite galleon]

Get the linkk

[tired kraken]

Then find the link for it and paste it into the colab

[finite galleon]

I thought julia gave it to you before

[brittle wing]

Iv sent the link

[finite galleon]

We touch a screen not electricity

[half cove]

Wtf am I doing its almost 3am and I've randomly went to get a glass of milk

[finite galleon]

Why milk

[half cove]

Idk 😭

[brittle wing]

Model 1

[brittle wing]

Model 2

[half cove]

You know what ima do it

[brittle wing]

These are all I need

[finite galleon]

Do what

[brittle wing]

I just need their vocals and the songs instrumental

[half cove]

The thing for him

[finite galleon]

Oh

[rare goblet]

Ayo? @finite galleon level 26 !!!

[finite galleon]

Gl man

[brittle wing]

Plz do it makes it easy for me and u bc u stay sane and I stay happy

[tired kraken]

I mean you should let him figure it out himself so he learns how to do it, but you do you

[pure scroll]

you want to seperate instrument and vocals?

[inland swift]

Hello

[brittle wing]

No

[inland swift]

I need help with making own ai voice

How i can make own ai voice

[brittle wing]

The ais sing the song but I need the instrumental from the song they sing

[violet heron]

-colab

[azure marsh]

Suggestions for @inland swift

☁️ Google Colabs

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

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• 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

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

[inland swift]

I have short samples

[violet heron]

RVC Disconnected

4th link

[wintry juniper]

hey im trying to train a model and it isnt working and all i hear is static when i use the model and on my tensorboard there are a bunch of wild triangles in the graph does anyone know what could be causing this so i can fix it

[inland swift]

I have colab already

[pure scroll]

record your voice properly with no background sounds

[inland swift]

How much samples do i need

[pure scroll]

is the dataset clean?

[inland swift]

What is dataset

[pure scroll]

3-5 min will be ok, try recording your voice reading something

[wintry juniper]

theres some weird parts in it where the audio goes a little weird but most of it is

[inland swift]

Wait a sec...

[pure scroll]

audios that we train through ai

[wintry juniper]

and theres no spaces in the audio or anything

[pure scroll]

cut it then!

[inland swift]

I thought i needed to record all words separately

[wintry juniper]

k

[inland swift]

I recorded all words separately and shortly

[pure scroll]

lol no!

[inland swift]

Bruh 😭

I need to re-record now

[wintry juniper]

could that really cause it to not even be able to speak though?

[pure scroll]

don't worry it happens

try reading a story or something ig

[violet heron]

No

[brittle wing]

Did u do the thing? Bc I need it rn

[inland swift]

I'm bad at using english language when i do speak

[rare goblet]

Ayo? @brittle wing level 5 !!!

[inland swift]

My accent is very bad

[half cove]

It's processing

[pure scroll]

if the audio that you train has instruments or background noise, static will be there on output audios

[rare goblet]

Ayo? @pure scroll level 16 !!!

[brittle wing]

Ok

[half cove]

Check bot spam

[inland swift]

Is it good if I'll just say 20 words?

[wintry juniper]

no i mean the WHOLE audio is static

it makes only static

[pure scroll]

record your audio in the language your are good at!

[finite galleon]

Every "bad" accent is a legendary accent in my eyes

[pure scroll]

and use the model in that language only

[inland swift]

Turkish and russian are my main languages irl

[pure scroll]

how many epochs did you trained it?

[inland swift]

Mostly russian

[wintry juniper]

112

[violet heron]

Was there any music in the background

[finite galleon]

How lmao

[wintry juniper]

no

[inland swift]

Idk what words i can say

[pure scroll]

record in russian only then

[wintry juniper]

idk 😭

[violet heron]

And did you lower the sample rate on colab

[inland swift]

Can i have your dm's?

[wintry juniper]

didnt use colab used applio

[inland swift]

And your dm's too

[violet heron]

Oh

[pure scroll]

there definetly might be background noises! try making the dataset clean and echo free as fck!

[finite galleon]

Nah i cant help you sorry

[inland swift]

You two seem nice to me

[inland swift]

I don't need help in dm's

[finite galleon]

I get insane when i help someone for some reason

[inland swift]

I just wanna be friends

[pure scroll]

use this channel only please, it helps us too (reputation maybe haha)

[inland swift]

Talk sometime etc

[inland swift]

Alright

[wintry juniper]

k but like shouldnt it atleast still produce something other than static?

[inland swift]

I hope I'll help people too in one day

[wintry juniper]

theres still a good chunk where its just talking normally so why cant the model even speak?

[pure scroll]

i'm sure the dataset is bad

[pure scroll]

[proud elbow]

https://discord.com/channels/1159260121998827560/1186762171204505801 may contain useful tips, and go figure

[inland swift]

This gives me ptsd

[wintry juniper]

maybe but idk i ran it through a bunch of stuff

[inland swift]

https://cdn.discordapp.com/emojis/1187548225956237332.webp?size=48

Embed fail

[pure scroll]

just make sure the person is speaking without having any echo's or artifacts

or try training higher too

400 epochs?

[wintry juniper]

ill try

[rare goblet]

Ayo? @wintry juniper level 2 !!!

[inland swift]

Any i want answer

[pure scroll]

check the graphs on tensorboard!

[inland swift]

How much words i need to record

Idk what to say tbh

[pure scroll]

https://tenor.com/view/embed-no-no-embed-megamind-megamind-meme-gif-25261934

[inland swift]

I need some list of words

[pure scroll]

embed fail me too

[inland swift]

Hahahah

[pure scroll]

read story maybe

[inland swift]

Let me do some list

Tea
Xbox live
Game
Bro
Sus
Player
Doing
Normal
Spectacular
Desperate
Lol
Joke
Joking

[inland swift]

Yo i have idea
I have suggestion for this server

[pure scroll]

server?

[inland swift]

I'm going to #1159516963014451302

Here
https://discord.com/channels/1159260121998827560/1191574043040678018

[half cove]

https://tenor.com/view/epic-embed-fail-ryan-gosling-cereal-embed-failure-laugh-at-this-user-gif-20627924

[inland swift]

I have idea about how i can make own ai voice

[rare goblet]

Ayo? @inland swift level 4 !!!

[inland swift]

Does someone have some samples?

[neat compass]

whats the best way to seperate vocals from lead vocals

[azure marsh]

Suggestion for @neat compass

Ultimate Vocal Remover

One of the best free and open source vocal and instrumental isolation tool.

🎄 Tip: Use /guides UVR to learn how to use it.

Ultimate Vocal Remover

This guide shows how to separate the vocals and the instrumental of songs using UVR locally
Link: https://docs.aihub.wtf/vocal-cloning-guide/vocal-and-music-separation/uvr

🎄 Credits: Nick088

Ilaria UVR

Link: https://docs.google.com/document/d/1M3rEj5RPSE60f17RdM9MaIKKkS2bLrj-dwwuCUOdRaY/edit?usp=drivesdk

🎄 Credits: Angetyde

[pure scroll]

@neat compass here you go

[neat compass]

Which model would be best to seperate lead vocals and backing vocals

[pure scroll]

MDX-net Karaoke 2 & VR-Arch 5_HP-Karaoke

[neat compass]

thanks

[molten pecan]

What's the diff between the two?

Which one is better?

[pure scroll]

ngl i prefer using 5_HP-Karoake
the 6_HP-Karoke is also gr8

[pure scroll]

the one that is newer is better

[molten pecan]

Gotcha thanks

[inland swift]

Guys is this normal when i do the colab thing

I'm not scared but still

[pure scroll]

yes, the colab do require access to your google drive

its normal and safe don't worry

[inland swift]

Alright

And i'm angry rn

When i'm trying to find my folder path it doesn't shows my dataset folder

Folders just disappear

How it works 😭

[pure scroll]

are you using RVC Disconnected one?

[inland swift]

What?

[pure scroll]

colab?

[inland swift]

I'm using website

[pure scroll]

screenshot?

[inland swift]

BRUH

WHY IT DOESN'T SHOW THE FOLDER

I'm trying to find dataset folder

[pure scroll]

drop the files here

[inland swift]

Nvm

I need to copy path's link right

Oops

I got confused

[pure scroll]

yes!

[inland swift]

I meant

I need to copy my zips path

[pure scroll]

yup

[inland swift]

Alright

I'm watching yt tutorial for that

I can send the link

Oh wait

I needed to make folder in dataset

I already regret for trying to make ai voice

It stressing

[inland swift]

So i click start at step 1 and after giving all needed perms
After i put the zip name in step 2
After i press start in step 3 after i'm opening the public url

After I'm going to training section in the website

If i fail i restart everything

Like i'm closing the url and reloading the colab website

[astral jungle]

Hello

[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

[inland swift]

my nerves are at my limit

i did 27 tries

it doesn't let me make my voice

and i'm mad

i even tried to do the
https://discord.com/channels/1159260121998827560/1160926856548200531

[proven hill]

NEW MOMMY VOICE MODEL!!!

Hi cutiepiessss you asked for a better mommy voice model for the voice changer and i listened!!!

Grab it here before it goes: https://ko-fi.com/s/98d841bf59

Also visit my shop for other voices like e-girl and v-tubers ones (free included!!)
https://ko-fi.com/ilariaowo/shop

i love youuuuuu

[glad zealot]

use !cp -f

[tepid orchid]

Oops wrong channel

[latent kettle]

Can I use UVR on Colab?

[pure scroll]

yes!

[latent kettle]

Please send me the notebook

[pure scroll]

https://colab.research.google.com/github/NaJeongMo/Colab-for-MDX_B/blob/main/MDX-Net_Colab.ipynb

[violet heron]

Bro

[pure scroll]

i saw that @violet heron

[violet heron]

Ilaria is deleting it

[pure scroll]

what did you do

[violet heron]

Tell her to use #1159290752195633273

https://tenor.com/view/paranoid-theyre-probably-watching-us-right-now-big-brother-is-watching-the-simpsons-bart-simpson-gif-5417245

[pure scroll]

for that merch thing she did?

[violet heron]

No embeds

[violet heron]

Yeah

[violet heron]

Say her name 3 times and she will appear

Ilaria

Ilaria

Ilaria

[latent kettle]

How to use this?

[violet heron]

read

[pure scroll]

https://docs.google.com/document/d/1HmkG9cmL8SLX7-vJcPT1-1KgUQtCrwXB8CicYmG4LW8/edit#heading=h.ub20t3fkdp39

[latent kettle]

Does it have DeEchoReverbe model? Czz i didn't find it in notebook

[azure marsh]

Documentation

📚 All-In-One English documentation

https://docs.aihub.wtf/

❔ English FAQ

https://docs.aihub.wtf/faq

🎄 Credits: AI HUB FRANCE

✨ More guides

Full AI Voice Model Training Guide (Local)

Link: YouTube
Credits: Christopher Villanueva

Model training with Mainline RVC

Link: Rentry
credits: Raven (ravencutie21)

AICoverGen Colab Guide

Link: Google Docs
Credits: Eddy (Spanish Helper)

Create a model with RVC disconnected (colab)

Link: Google Docs
Credits: Angetyde

How To Make an AI Cover With Ilaria RVC

Link: Rentry
Credits: 👽 Julia (ailen2091)

[ocean horizon]

how do i hear myself while using a virtual audio cable to output the sound?

[fossil turret]

hey i have some problem so i use rvc colab version and i'm running into some problem, when there is no input the "res" are running at 200 avg, but when i talk or there is an input it start going crazy like 20000 or even 30000 so i need a solution please my internet speed are running 45 MB download and 30 MB on upload

[glad zealot]

what is the chunk

[fossil turret]

320

it's running normal to at passthru

[rare goblet]

Ayo? @fossil turret level 1 !!!

[fossil turret]

i also have tried to change every single setting but none of them works

[plush compass]

What INDEX_RATE do you all set when making songs? In the case of an AI model of an actual singer, not an animation or game character.

[blazing garnet]

What do you use?

What software or colab or what?

[plush compass]

I used 0 or 0.05, but I just set INDEX_RATE to 1 and made a song, and it came out better than I expected, so I'm curious about what other people think.

colab

[pure scroll]

many people use the default index setting

if changing it helps you

gg for you

[plush compass]

thx

[blazing garnet]

Use 0.75 it's the default but I may have been late

[inland swift]

Hello

[rare goblet]

Ayo? @inland swift level 5 !!!

[inland swift]

I'm having problem with index thing

For some reason i can't make index file it gives me error

Is index file necessary?

This chat is dead 💀

[flint kernel]

fr

[inland swift]

I need answer

[flint kernel]

oh

yeah

I believe so

[eager pelican]

its literally named train index
when you train your model
just train index along
then creates a index file in the logs folder and done

if there's a technical issue literally give the console log 🔫

[inland swift]

I mean

I did training but didn't got index file

After training i pressed into index thing

To create a index file

I don't have index file anywhere

[flint kernel]

why is my shit cutting out at the end :'[

[rare goblet]

Ayo? @flint kernel level 1 !!!

[inland swift]

Swears

https://cdn.discordapp.com/emojis/1187548225956237332.webp?size=48

[flint kernel]

🤓

[eager pelican]

there are seperate buttons for train index and train model

[eager pelican]

fake nitro moment

[flint kernel]

so yk

why is my model cutting out at the last second

[eager pelican]

fym

[inland swift]

Is index even necessary?

[flint kernel]

it used to fine

fym?

[eager pelican]

depends on how the model is
if its bad index can help it
but if its really bad index makes it worse
but if its really good index isnt needed and often just adds noise in my experience
soooooooooooooo uwu

[eager pelican]

"fuck do you mean"

[flint kernel]

o

LOL

[eager pelican]

wdym by the model is cutting at the end

[flint kernel]

so basically

the last 2 seconds

it just cuts it off

[eager pelican]

when you infer?

[flint kernel]

no I mean like

[eager pelican]

like input an audio and then changes the voice

[flint kernel]

whilst using the voice changer

[eager pelican]

oop uwu
#🔍│help-w-okada

realtime voice changer do be on tehre

[flint kernel]

oopsies

[eager pelican]

dw the chat is already dead

the server is dead

smh people got lives and schools

[flint kernel]

you should help me fr

[eager pelican]

i dont use realtime :3

[flint kernel]

flip you!!!

:'[

[eager pelican]

sorry that im fine with not being a cat girl on the internet cat fishing guys

[flint kernel]

:'3

https://media.discordapp.net/attachments/1151870151105450065/1180310107851067462/furry_dance.gif?ex=657cf472&is=656a7f72&hm=10a17ce64a001ae19b2161c872a0e898cb86054522569caa52de0f6e7b3d4147&

bruh

🙄

discord moment

[eager pelican]

smh

[flint kernel]

smh

[left oasis]

Why wasn't my model posted?

[inland swift]

Never uwu at me ever again pls

[eager pelican]

uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu

[flint kernel]

uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu uwu

[eager pelican]

when you submit a model in #1159514067187277865
when it gets accepted
you get model maker role
then you post your model manually in the #1175430844685484042 channel

but you dont have model maker role
so ig it wasn't approved

[eager pelican]

your mom

this is a help chat

not joke :3

[inland swift]

Saying about mom is weak thing to do

[ancient flower]

Trying to make this one song with the fnaf 4 vocals from TLT, but the result keeps spitting this shit out no matter what settings I'm messing with it.
It makes it sound like incoherent whispering (I'm using CoverGen_NO_UI collab)

[left oasis]

It's just that they didn't even write to me how can I know

[eager pelican]

what model did you use for isonlating it

[eager pelican]

oh it takes a while
Wait for now
they do be busy with hundreds of submissions

[ancient flower]

old model I made ~6 months ago https://huggingface.co/filler5736/mayamodelNEW250/tree/main

wait fuck

[left oasis]

oh ok

[eager pelican]

no i mean for the UVR
what model is it

VocFT
MDX23C
InstHQ

[sand iris]

@eager pelican how do ai cover ?

[eager pelican]

first you go and touch grass

when you do that come back to me

[ancient flower]

For getting the vocal? I just got it from the DL link TLT provided and applied deEcho-deReverb on UVR, which I guess comes from VR Architecture if that's what you're wondering

[eager pelican]

so it was already without instrumentals in the first place?

[ancient flower]

yea

[eager pelican]

so you want to de reverb

for de reverb ye VR arch good

[ancient flower]

I already de reverbed it

[eager pelican]

then why does it sound bad

[ancient flower]

that's what I want to kno

[eager pelican]

send me the notebook link

lemme check what model it uses

[ancient flower]

https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk#scrollTo=DtHla7cvRRqh

[eager pelican]

ok wait for now

[eager pelican]

isnt this just for cover gen
this isnt for de reverbing apparently

or i may be wrong obviously

[ancient flower]

yeah this is just for ai cover generating

[eager pelican]

the only thing i see is

# @markdown Reverb Control
REVERB_SIZE = 0.15 # @param {type:"number"}
REVERB_WETNESS = 0.2 # @param {type:"number"}
REVERB_DRYNESS = 0.8 # @param {type:"number"}
REVERB_DAMPING = 0.7 # @param {type:"number"}``` which i dont see any model name so idk if its just adding reverb or removing reverb

[eager pelican]

so for de reverbing you need another notebook

there's this if you want https://discord.com/channels/1159260121998827560/1159588254220288121

[ancient flower]

no I don't need to de reverb, the issue is that my outputs are sounding like weird whispering for some reason, like it's not trying to convert the whole vocal track into my ai voice for some reason

[eager pelican]

then its prob jsut a bad voice model or bad input audio

or bad settings which i'd say prob not

[ancient flower]

the og voice from the song is already super synthesized sounding, so maybe the notebook is confusing it for instruments?

[eager pelican]

then its prob a bad input
make sure the input atleast sounds human
but RVC can prob ignore that and still give a better output
so my bet is that its a bad model

give me your input file

[ancient flower]

Like here or by DMing it to you? Since I know this place is a bit on lockdown from sharing vocals and instrumentals of copyrighted songs and whatever

[eager pelican]

DM would be better

why would you even give instrumentals as input for RVC

you need to isolate the vocals and then input that

@ancient flower ok ya its kinda a bad input i'd say
but you cant really fix that and i'd say RVC can ignore it mostly
so its just a bad model

[ancient flower]

Okay, also I was just saying sharing instrumentals and vocals are forbidden here and I was not inputting instrumentals into RVC, or whatever. But Thank you

[eager pelican]

reasonable

atleast bro read the #📜│rules fr

[pure scroll]

yeah our server got deleted cuz of copyright once before

[forest gate]

How do i download voices from kits.ai? im very new to everything, mentioned website included

[pure scroll]

honestly kits.ai is trash

[eager pelican]

download from https://weights.gg/

[pure scroll]

THIS IS THE WAY

https://tenor.com/view/literally-rei-chico-lachowski-chico-rei-gif-15056972337014391696

[eager pelican]

ahem

people are sleep no shouting

[pure scroll]

i need embed perms

[eager pelican]

get rekt

[pure scroll]

aaaaaaaaahhhhh

[pure scroll]

what's that

[eager pelican]

kits.ai is dead
and its really bad and propietory so to say

so just either use #1175430844685484042 or https://weights.gg/

[proper shale]

https://tenor.com/view/holy-moly-holy-moly-gif-25041694

lusbert helping in help-rvc is real now

[forest gate]

oh ok thnx

[eager pelican]

always had just that i couldnt care less so i stopped dealing with trolls

past server i had 10k messages in here

[main trail]

Is rvc better than w-okada

@eager pelican

[eager pelican]

w-okada and RVC are KINDA the same thing
just that Okada is for realtime usage

[proper shale]

...W-Okada uses RVC?

[proper shale]

yeah what bro said

[main trail]

Also i have a problem where it sounds really good but itll randomy cut out or muffle a ton is there a fix?

[eager pelican]

or does it
its a secret tech
the goverment is hiding it from us

[eager pelican]

literally there are guides for voice changer

https://rentry.co/VoiceChangerGuide

smh

[eager pelican]

nuh uh

[finite galleon]

Rvc is rvc
Wokada can use rvc

Rvc is for audio files
Wokada is for real time audio

[proper shale]

Do you have a stationary mic?

Because it could help to stop the muffling

[eager pelican]

shocker
crazy ik right?

[finite galleon]

[limber flume]

I want to make an AI cover for "You're A Mean One Mr.Grinch" with Tyler the creators voice but different lyrics, how do I do this?

[proper shale]

you need to record new lyrics for that then

or get someone to do that

then AI it and put it back with the instrumental

[eager pelican]

get lyrics
seperate instruments
get smth like XTTS
input a 5 second part ofthe voices from the song into XTTS
make it read the lyrics
and then run the output through RVC
since XTTS can clone voices
and then you can make the cloning better by running the TTS output through RVC

[limber flume]

Respectfully.

[eager pelican]

so jsut make XTTS sing the lyrics and then run the output through RVC and then put vocals on instruments and done

[limber flume]

What the fuck

[eager pelican]

https://huggingface.co/coqui/XTTS-v2

this can clone voices
its meh
But RVC exists for a reason
it will make it way better

[proper shale]

bruh nah

it probably

[eager pelican]

ig yall are lazy

[proper shale]

will not sound great

for singing

all I'm saying

[eager pelican]

it will sound better than just a generic tts sound

[proper shale]

anyone can sing lyrics, or maybe they can get a friend to do so

[limber flume]

Could I pay someone 30$ to do this? Cuz Im outta town

[rare goblet]

Ayo? @limber flume level 2 !!!

[eager pelican]

yk... if you want you can pay them for an already made easy gui and then use that
or just write your own code to make it do that
but it can :3

or jsut use diff singer

[limber flume]

And this sounds too complex

[obtuse otter]

hey im using an rvc and the voice/quality seems fine, but for some reason when i speak the audio sounds distorted/muffled for a few seconds and then goes back to sounding normal

[eager pelican]

not anyone that i know of

[eager pelican]

or well ya that'd be much better

[obtuse otter]

is there a way to fix that?

[proper shale]

It really isn't tbh, just gotta get used to it

[eager pelican]

@limber flume either sing the lyrics yourself and then run RVC on you singing
or use some TTS

[limber flume]

K

[eager pelican]

idk about diffsinger since its mostly for vocaloids
but diffsinger sounds promising
https://github.com/MoonInTheRiver/DiffSinger
i havent heard of any one use it tho
only 1 person

[obtuse otter]

its only with this voice that im having this problem

[eager pelican]

it got TTS support and shit

[obtuse otter]

does it have something to do with my mic quality?

[eager pelican]

prob a bad voice model

[eager pelican]

also yea

[proper shale]

^

[eager pelican]

RVC depends on two things
voice model and audio input

[obtuse otter]

its weird tho cause the voice sounds really good when its working

[eager pelican]

from RVC?

[obtuse otter]

yeah

[eager pelican]

so on a normal input it sounds good?

[obtuse otter]

i think its an 800 epochs one

[eager pelican]

and on your mic it sounds bad

so its prob jsut a bad model

[obtuse otter]

dang

[eager pelican]

epochs mean nothing in voice models so to say
it heavily depends on batch size and step count and dataset

epochs have no meaning in terms of quality

[obtuse otter]

tryna sounds like carl from jimmy neutron btw

[eager pelican]

higher epoch doesnt mean better model uwu

[proper shale]

they do mean something

just not that it's high quality yeah

[proper shale]

closer to dataset :)

[obtuse otter]

im an ai noob so still tryna figure this stuff out

[rare goblet]

Ayo? @obtuse otter level 1 !!!

[proper shale]

if the dataset sucks model sucks

[obtuse otter]

shieet okay

[eager pelican]

epochs are just how much it reads through it
the only effect it will have is underfitting and over fitting

[obtuse otter]

thanks guys 🙂

[eager pelican]

and if the dataset is good
the model can still suck
As i said
it depends on step count and batch size

[proper shale]

you're welcome, let us know if you've found a better voice model or something

[eager pelican]

if i do batch size 1
the model would be the worst thing to ever exist basically

[obtuse otter]

bet

[eager pelican]

cause it tries to replicate the dataset 1:1

which obv
no dataset is that perfect

never will be

[blazing garnet]

.

[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

[drowsy solstice]

wont it get disconnected after a while?

[latent kettle]

It will not disconnect you, but you have to avoid inactivity

[quartz frost]

i need a favor

not like technical help

but a helper to do them for me

is there a helper who can ai convert for me?

also a guy is singing the song and the ai is a girl so can they also manually set it up

[pure scroll]

what

[quartz frost]

i got the ai voice model

but if i were to convert when the singer is a guy

it would be, broken you know

so theyll need to set the pitch and everything

[pure scroll]

just use the hit and trial method bro

try moving the pitch to + to sound it female
to - to make it sound male

[quartz frost]

its a favor, bro

im asking if someone minds doing it for me

for now

[pure scroll]

can i? but am not a helper so......?

[quartz frost]

anyone, as long as anyone is willing

alright thank you btw

[pure scroll]

aight send it to me, i'll try changing it

[quartz frost]

👍

let me set it up in huggingface for you

[errant thistle]

i having problems finding wordband bot does anyone know how to access it?

[left oasis]

what is it?

I trained the model and all the ticks

[pure scroll]

read "exception"!!!

[left oasis]

i read

And what does that mean

[pure scroll]

did you run any option saying 'Feature Extraction"?

[left oasis]

yes

[pure scroll]

try not running it!

restart the runtime

and not run it

[proper shale]

wha

[left oasis]

Do I have to wait again?

[proper shale]

no

that's a dataset problem

[proper shale]

they might have formatted the dataset in a wrong way

[pure scroll]

oh

[errant thistle]

i having problems finding wordband bot does anyone know how to access it?

[left oasis]

I did everything as always

[proper shale]

maybe the audio you're using is problematic

[left oasis]

It lasts 10 minutes and it's clean

[proper shale]

how did you export it tho

[left oasis]

mp3

[left oasis]

I'm always in mp3

[proper shale]

where exactly

cuz i recommend using wav to export

inside audacity

[left oasis]

wav?

[proper shale]

yes, exporting as wav is always better

it prevents errors

[left oasis]

holy ahahha thanks

[left oasis]

It's strange, I did everything well that time.

[proven hill]

NEW MOMMY VOICE MODEL!!!

Hi cutiepiessss you asked for a better mommy voice model for the voice changer and i listened!!!

Grab it here before it goes: https://ko-fi.com/s/98d841bf59

Also visit my shop for other voices like e-girl and v-tubers ones (free included!!)
https://ko-fi.com/ilariaowo/shop

i love youuuuuu

[inland swift]

what

how it does work bruh

i even did rename

it makes no sense

i'll reload the website

[proper shale]

did u allow drive to connect

[proper shale]

uhhhh i think u didn't run the dependencies stuff correctly

[inland swift]

yeah

[inland swift]

i did

[proper shale]

disconnect and restart runtime just in case

and make sure its using ur google acc

[inland swift]

i'll reload website

ok?

[proper shale]

nope, you have to restart the runtime n stuff

[inland swift]

how to restart

i never used this website before

[proper shale]

see the top buttons? there's a runtime button there

[inland swift]

what this needed for

[rare goblet]

Ayo? @inland swift level 6 !!!

[proper shale]

ofc it's probably in a diff language

[proper shale]

oh this is essential lol

[inland swift]

i speak russian

[proper shale]

thats why nothing's working

[inland swift]

i also know turkish

[proper shale]

yeah just pick the 5th option up there

after file and stuff

and then disconnect and delete runtime

something similar to that

[proper shale]

also i recommend minimizing the Dependencies step

[inland swift]

how to disconnect

[inland swift]

bruh i'm confused

it's hard

[proper shale]

again, the runtime button

[inland swift]

alright

[proper shale]

there should be a "disconnect and delete runtime" thing there

or something similar in ur language

[inland swift]

there's no disconnect

[pure scroll]

look at top right

[inland swift]

this thing worked

[proper shale]

it's the ninth option

and then minimize the dependencies step so u run everything there at once

no need to do manually

[inland swift]

still confused

i'm trying to process the information rn

[inland swift]

i got into this point

[proper shale]

oops, space in the name of the model

cant have spaces

[pure scroll]

is he training model?

[proper shale]

try doing everything from set training variables to the training model again

[inland swift]

i have model which is named
Russian RasuL

[proper shale]

but set a different one

name it just RasuL or Russian_RasuL

[pure scroll]

use the RVC Disconnected Colab

[inland swift]

oh ok

[proper shale]

they are

[inland swift]

i'll reload the website

[pure scroll]

hmm

[proper shale]

no need

just set a diff name in set train variables

and do the preprocessing and feature extraction again

[inland swift]

already reloaded

[proper shale]

don't do all the steps again, just do the ones i said

u should be fine

[inland swift]

this time i didn't use any space

i used _

and it still does give error

i'll try without using _

[proper shale]

because u gotta do all the preprocessing/feature extract stuff again

after u set the new trainin variable

[inland swift]

man

this sucks

[half cove]

Idk if I'm dumb but training index on colab never works for me 😭

[inland swift]

did already

[proper shale]

ur not doing the dataset formatting correctly

probably

[half cove]

Like file format?

[inland swift]

but it still doesn't let me

[proper shale]

no as in

folder > audi

auidos

audios

[inland swift]

maybe i should show screenshare?

[half cove]

Wat

[proper shale]

cant call

[eager pelican]

if he can train a voice model
why wouldt he be able to train a index ;3

[inland swift]

in level 5 vc

[proper shale]

u need to put ur audios in a folder, make sure folder has the same training variable name

[proper shale]

i literally cant

[inland swift]

ok

[eager pelican]

@half cove can you even train your voice model normally in colab
or just can train but index wont work

[brittle wing]

It's been 2 days where is my ais

I need my second voice for the song

[inland swift]

i don't understand what this thing does bruh

[half cove]

Index

[eager pelican]

@proper shale see?
he can train normally so his folder structure is fine

[brittle wing]

@half cove by any chance could u do that thing for me again but with a different voice model bc the ai is still broken

[eager pelican]

colab issue smh

[half cove]

Im learning colab so i can help people in it

[inland swift]

same

[brittle wing]

model:joe_hawley_tally_hall youtube:https://youtu.be/rfUeWe7u364?si=ji7PGT8Yokov37lb

@half cove can u make the ai generate this

[eager pelican]

right

[brittle wing]

Bc it's voice ai is broke

[eager pelican]

@half cove why do you even want index
index is useless imo ;3
just use without index

[half cove]

So I can help people use it

[eager pelican]

whats the error you get when trianing index

[half cove]

That I didn't run feature extraction

[eager pelican]

bruh

then do that? duh

wait how did you train a voice model if so

[eager pelican]

huh 😭

[half cove]

I have

[eager pelican]

@half cove give me the link to the notebook rq

[eager pelican]

then its just bad code

[half cove]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• 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

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

[eager pelican]

which one

[half cove]

Disconnected

[eager pelican]

let's see

me waiting 2 hours for page to load

[eager pelican]

did you

Save preprocessed dataset files to Google Drive

[half cove]

Yes

[eager pelican]

you fully did feature extraction as well?

[half cove]

I'm not a idiot

[eager pelican]

idk but
"Note: It's rare, but you may encounter a bug with this cell that requires a runtime restart. If this happens, restart the runtime, re-run the "Set Training Variables" cell, then re-run this cell."

[half cove]

Ofc I did

[eager pelican]

did you read the console log tho uwu

[half cove]

Yea that never works

[eager pelican]

lol

[half cove]

Truly a skill issue

[eager pelican]

RVC does have a ipynb included with its files....

so basiclaly a notebook

[half cove]

I mean I can at least help people with the other bits

[eager pelican]

i wonder why no one uses that

does the console even say anything when it doesnt train index

does it even HAVE a console

[half cove]

No cuz it's colab

[eager pelican]

it prob does

[half cove]

I could just do it locally but most people use colab so that wouldnt help me help people

[eager pelican]

make your own notebook

[half cove]

I'll figure it out later

[half cove]

Hmm

[eager pelican]

uwu do it

[half cove]

Time to steal code

[proper shale]

bro...

it says no-feature-todo

[eager pelican]

enjoy having code that wont work cause its a mess

[eager pelican]

lmfao

[proper shale]

ur audio is screwed

or how you zipped it was wrong

[eager pelican]

@half cove what did i say
read the godamn console

[proper shale]

thats always the main reason

[half cove]

It's colab

[proper shale]

there's a easy log to read bro

it said "no-f0-todo"

and that model is probably fucked

[eager pelican]

either the config for the thing has bad setting which i doubt
or that his audio is so bad that it didnt grab anything

[half cove]

Nuh uh

[eager pelican]

bozo listen to mj

[half cove]

It never said that

[eager pelican]

you just didnt see it

cope

[brittle wing]

How do I fix this iv logged into other emails on the website

[eager pelican]

@half cove give me your log file

i wanna read it

i need to read it actually

[half cove]

On mobile rn I ain't doin it again on mobile 😭 it's prob just that I did something dumb

[proper shale]

honestly bro, just use ilaria rvc

[eager pelican]

yea

[proper shale]

for the time u are waiting it'd be better

heck even AICoverGen

[eager pelican]

@proper shale do you have his logfile cause apparently you just read it for him

[proper shale]

-colab

[azure marsh]

☁️ Google Colabs

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

🤗 Hugginface Spaces

• Ilaria RVC, by thestingerx Hugginface Spaces
• 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

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

[eager pelican]

@half cove when you get home gimme your log file 🔫

[half cove]

[eager pelican]

never use colab on a fricking phone

never...

[proper shale]

• u gotta zip a folder with audio

not just audio

lol

[eager pelican]

lol he didnt even read the instructions?

[half cove]

I did that

[proper shale]

ok but how was the audio exported

[eager pelican]

the same as the instructions say?

[eager pelican]

cause it exports it...

how else do you think it usest he files

[proper shale]

and was the folder name the same as the training variable

[proper shale]

nah I mean like where did he get the audios from

[half cove]

That's prob it

[eager pelican]

reasonable

[proper shale]

see

[eager pelican]

yea

[proper shale]

¯_(ツ)_/¯

[half cove]

I said that before I swear

[eager pelican]

nuh uh

[half cove]

Leme see

[eager pelican]

just cope and redo it

[half cove]

[toxic maple]

When I'm training a pth file, I'm getting this error "训练结束, 您可查看控制台训练日志或实验文件夹下的train.log", how can I fix this?

[proper shale]

check command line

see if there's some error

[toxic maple]

It says training echo 1, but it stops while it is 0% finished

after around 100 seconds

[proper shale]

have you done all the previous steps (preprocessing, feature extraction)? and what is your gpu?

[toxic maple]

yes, I have a RTX 2070 (in a laptop)

[rare goblet]

Ayo? @toxic maple level 1 !!!

[toxic maple]

cause every step went good, until this last step

[proper shale]

are you sure? how are the logs in preprocessing and feature extracting tho?

[toxic maple]

It creates an index file, but then stops.

[proper shale]

u can copy and paste

[toxic maple]

after the first step: end preprocess

[proper shale]

just start and end?

so preprocess ain't the problem, most likely

[toxic maple]

do u want the cmd logs?

[proper shale]

yeah sure

[inland swift]

JUST WHY THE TRAINING DOESN'T WORKS

I didn't anything correctly bruh

Did*

[inland swift]

I hate autocorrect

[inland swift]

How it can't find if i literally put names correctly

[proper shale]

again, running the dependencies all at once prevents errors

[inland swift]

No space etc

[inland swift]

I tried

[proper shale]

ur running each step in there manually