✨│ai-help

[quaint dock]

okay, can I delete the rest?

[half cove]

Don't u have index?

[quaint dock]

I have the index

[half cove]

U can del the d and g pth of u want

[quaint dock]

alright, thank you!

[half cove]

Only used for training

[wintry adder]

i keep getting this error when i tried to train a model: torch.foreach_add(device_exp_avgs, device_grads, alpha=1 - beta1)
RuntimeError: The size of tensor a (24) must match the size of tensor b (16) at non-singleton dimension 2

[clever valve]

@proper shale I got the vb thing but when it’s input on discord it plays no sound on discord

[dense pivot]

1.5 - 2 hours I can confirm, I'm using 3070 as well

[vapid gust]

I’ve got a question

Why does my voice covering always take the accent of the input audio rather than the model accent

The accent on the input audio is always much heavier on the output

[molten pecan]

You need to tweak the search feature ratio

You know what that does?

[vapid gust]

No

[vapid gust]

I’ll put it on 0.99

[molten pecan]

Let me explain

[vapid gust]

Isn’t it just more of the model

[molten pecan]

Yeah

[vapid gust]

Being used

Ye ok

I’ll try 0.99

[molten pecan]

More of the characteristics of the model

Sure

[vapid gust]

Does 1 work

[molten pecan]

Just play around with it until you find the sweet spot

Of course

[vapid gust]

Is there a difference between

0.98 and 1

[molten pecan]

Yeah but subtle

[vapid gust]

ah ok

[sick comet]

what is rvc

[rare goblet]

Ayo? @sick comet level 1 !!!

[sick comet]

oh

can someone help with searching for models every time i search for a new one like it has the same voice

[violet heron]

Ok

That is kits.ai

Which is bad and stupid and terrible

[sick comet]

oh

then how do i find another site

[violet heron]

You cannot download models from kits.ai

[sick comet]

cuz all of the voices were taking me ther

er

like every time i search it took me there

[violet heron]

Use weights.gg to find models

Or search using @shy zealot instead

[sick comet]

u r my savior

[violet heron]

Or in #1175430844685484042

[sick comet]

i try but

it only goes back to messages like 2 days ago

[violet heron]

Those are reuploaded models

[sick comet]

is the server just rlly new

oh

[violet heron]

A mod accidentally deleted the channel

So we had to use bots to reupload everything

[sick comet]

ohhh i see and this was recent then?

oh

[violet heron]

weights.gg should have about everything

[sick comet]

uh

i have to ask how do i look at the audio samples

cuz im looking at this but it doesnt sound like her and it said something ab diverse range so i thought thered be like an audio sample with the thing on

i think it just was bugged like the site

cuz it sounds diff now

[violet heron]

There should be a bunch of them below right?

[sick comet]

yes

[rare goblet]

Ayo? @sick comet level 2 !!!

[sick comet]

are they for the voice changer or just audio samples

cuz im using it now and none of them rlly sounded like this

[violet heron]

Those are samples

[sick comet]

wait they do now

i swear they sounded different

tysm

[cursive vault]

im gonna try install applio locally

[cursive vault]

ok it didnt work

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[copper pebble]

ty

[dense pivot]

can I watch youtube when my GPU training the model?

[hot lagoon]

How do I apply to be a Model Master?

[south vine]

what's the best CPU trainer?

i seem to have ran out of credits on colab somehow

[frigid sapphire]

if you have a credit card and a gmail account, you can take a GPU on Google Cloud and connect it to Colab.

here's a quick guide: https://black.dubverse.ai/p/running-rvc-models-on-the-easy-gui

Feel free to DM incase you get struck anywhere.

[sand iris]

You can also (ab)use the free tier, but the UI isn't as comfy

[brittle wing]

Where can I download tensorboard?

[last patio]

Any tip to avoid my model sounding like a heavy smoker?

[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

[median agate]

can't resume my model on rvcDisconnect (colab rvc)

[unreal lichen]

attributeerror: 'nonetype' object has no attribute 'dtype' rvc how i can fix this with rvcweb gui v6

[copper pebble]

https://docs.aihub.wtf/guide-to-create-a-model/tensorboard-rvc

i thinl this should help u

[tough crane]

At what point do I stop training? [Like 2k, 5k, etc..] (I can't read graphs)

[copper pebble]

@proper shale 9k?? (1st one)

[median agate]

maybe here?

[tough crane]

Is "d/total" actually useless? In the previous guide on the OLD A.I. Hub, I see a random guide that says only "g/total" is very important in making models. Now that I read on the new guide, I'm kinda confused.

[tough crane]

I don't have a savepoint (IDK what it's called) around the 9k area.

[tame mica]

d and g is both important i believe

[tame mica]

checkpoint

[copper pebble]

ofc

[tough crane]

I kinda save the pth file around the 16k part only sadly

[austere isle]

what does "res" means in rvc?

[rare goblet]

Ayo? @austere isle level 1 !!!

[glad zealot]

probably response

response time?

[tough crane]

Could you explain why I should stop training around the 9k part (1st graph)?

[copper pebble]

im not that good but i think its cuz after 9k steps it start to go a lot down so i thought about overtraining

thats why i pinged MJ cuz im not really sure

how much epochs did u put and how much was ur dataset long?

[tough crane]

15:08.555 of dataset (singing only; MM:SS:ms)

1 epoch, I guess (if I reach at 200 epoch checkpoint)
I try to train my model to stop at 1000 epochs.

[copper pebble]

i see

i dont really know how resume works

but

[proper shale]

still going

[copper pebble]

k theres the master

ahaha

ty MJ

[proper shale]

nawww lol

[tough crane]

Colab limit reached lol

[proper shale]

keep going afterwards though

[tough crane]

ok, tnx!

[rare goblet]

Ayo? @tough crane level 3 !!!

[proper shale]

maybe give us kl and mel too

[proper shale]

np ;)

[tough crane]

ok, tomorrow I guess! (Unless my family need me to help with the family business)

[proper shale]

ight

good luck

[distant turtle]

best method to remove dirt from audio and echo/reverb, I have already tried uvr5 and it worked partially but some reverb still remains, then I cleaned the audio with adobe podcast but the audio makes it unnatural, do you know any other methods?

[rare goblet]

Ayo? @distant turtle level 4 !!!

[rare goblet]

Ayo? @brittle wing level 1 !!!

[silver anvil]

you mean okada?

a realtime voice changer?

#🔍│help-w-okada

what?

@proper shale can u help her?

im lost

[molten pecan]

Is voice.ai for converting your voice in real time?

[silver anvil]

omg julia is here

[molten pecan]

Like changing your voice in calls and stuff

A

[molten pecan]

Yeah I usually go here I like helping people >.<

You made an AI cover?

I see

Let me explain

There are two types of ways to convert your voice using AI

One is by inference, which is converting audio <files>, not for calls

And the other is converting in real-time, for calls

We recommend using w-okada since it's free and unlimited

There should be a guide on how to use it on #1159513888199540817

Millions of normal people and developers have been using it for months, if it had one we would have known at this point

Plus it's open-source, you can check the code by yourself if there's anything weird

You can always run a windows defender scan though

If you need any help ask on #🔍│help-w-okada

I'm not familiar with okada unfortunately

Good luck!

Like I said, I'm not familiar with okada so I don't know how to help

Ask on #🔍│help-w-okada

Good luck!

[mint nimbus]

Guys, I have problem

[rare goblet]

Ayo? @mint nimbus level 1 !!!

[mint nimbus]

wait web server

When I launched programm frist, it worked greatly

Now I try to launch and problems with web server

[proper shale]

Do you have an antivirus/firewall activated?

[mint nimbus]

No

[proper shale]

Delete stored_setting.json and restart W-Okada. Should work after that!

[mint nimbus]

It worked! Thanks a lot

[proper shale]

Np ;)

[oak finch]

Heyy yoo

Is there any platform except google collab to run Realtime RVC?

Cz it keeps getting disconnected

[proper shale]

U can run it locally if you have a good GPU

[grim bay]

How to fix this please. I'm using this
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS

[glad zealot]

do disconnect and delete runtime and run the cells again since you probably missed a cell

[proper shale]

^

also run all the dependencies steps together

because uh it wont work otherwise.

[mint nimbus]

My disk C is full

is this program to blame?

And how can I clean data files of this programm?

[violet heron]

Delete it

[proper shale]

occupies 5 GB of space maximum

so no

[mint nimbus]

I installed programm on another disc

[rare goblet]

Ayo? @mint nimbus level 2 !!!

[mint nimbus]

Maybe it also work with C

[proper shale]

I'm not sure if it works on another drive

But try it

[mint nimbus]

Ok

Is it neccessary to hold programm on SSD?

Or it don`t affects perfomance?

[proper shale]

I don't think it affects it that much

I'd say, try it

[grim bay]

What is the next step after trained?

[proper shale]

Now Export model

Have you trained Index yet though?

If not, train Index first then export model.

[grim bay]

This one right?

[proper shale]

yes

[grim bay]

Uh....

I did run Feature Extraction

[proper shale]

yeah your training screwed up..

how does the feature extraction step look?

[proper shale]

yep... didnt even train

how'd you export the dataset?

[grim bay]

I don't know, it's my first time using it

[proper shale]

what's the format of the dataset? where did you get it from?

[grim bay]

I use my own wav dataset

[proper shale]

Where did you export it from

Audacity?

[grim bay]

I use UVR to extract the voices

I'm not even use Audacity

[proper shale]

You definitely should

To trim silences n stuff

[grim bay]

Can you help me take a look at my dataset?

[proper shale]

im a bit busy so not really

[grim bay]

How about I just send you and you can take a look at anytime

[proper shale]

You can do that yourself honestly

I think you should just truncate silences and noise gate on Audacity, then export it to wav and see if it works

[grim bay]

Btw, is there a tutorial for the RVC V2 Disconnected?

[carmine hearth]

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

[proper shale]

yup

[glad zealot]

where there is?

how long have i been here? XD

[barren hornet]

better to have many different audios

it will work with one big, but it's better to have many small

also ensure that your audio(-s) have no breath, laugh and any other non-voice sounds

[grim bay]

I didn't find the .index files in here

Am I still doing wrong?

[glad zealot]

is that logs?

its been so long since i last use disconnected so i forgot stuff already

[grim bay]

It's in the Google Drive

[barren hornet]

It's better to have different voice expressions, but since its robotic, I think it will be fine

[rare goblet]

Ayo? @barren hornet level 1 !!!

[half cove]

did you train index

[grim bay]

After I extract from notebook to Drive

[grim bay]

Oh...It doesn't work

[half cove]

what do you mean?

[grim bay]

It shows like this

[half cove]

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.

[grim bay]

Still doesn't work either

[pseudo flint]

Guys I have 4070 ti but I only get 1 epoch per 1.30 minutes

why so slow?

[dusty rune]

guys how do i use Harry's RVC Dataset Maker

[grim bay]

Please help me fix this

[feral gale]

does applio work at all

i was on OG ai hub and just found out about here so

sorry if im outta the loop

[glad zealot]

yes

it works

[feral gale]

does it work with paperspace? thats what i use

[glad zealot]

no idea

[feral gale]

hmmm

okay

[glad zealot]

i only make a version for colab

made

[feral gale]

does it work without ppl getting banned

[glad zealot]

havent heard anyone getting banned ngl

so should be good

[feral gale]

on old colab u got banned ☠️ thanks tho im gonna look into this

do u have a link for the colab?

[glad zealot]

the old colab doesnt work anymore

its more like broken than banned

[feral gale]

good to know

[molten pecan]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Suggestions for @feral gale

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[molten pecan]

Oops

-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]

There we go

[feral gale]

thx sm

[rare goblet]

Ayo? @feral gale level 1 !!!

[uneven holly]

I'm using the colab thing to make a voice model but it loses connection so fast im not able to save the voice

What should I do

[violet heron]

Which colab are you using?

[barren hornet]

i had model maker role on old server, but old server exploded

[uneven holly]

Retrieval_based_Voice_Convertion_WebUI.ipynb

[uneven holly]

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

My voice didn't even trained yet this time

[violet heron]

That colab doesn’t work anymore

-colab RVC Disconnected USE A ALT ACCOUNT

[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

[uneven holly]

Idk it works but a little bit

[uneven holly]

So whats the best colab to do voice models

[violet heron]

Use or make?

[uneven holly]

Make

[violet heron]

RVC Disconnected

-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

[violet heron]

2nd link

[mint nimbus]

latest don`t work

[violet heron]

USE A ALT ACCOUNT

[uneven holly]

OK ty

[violet heron]

Are you on AMD or NVidia

[mint nimbus]

nvidia

[uneven holly]

where is says load dataset should i archive the voice that i want to train?

[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

[uneven charm]

I'm not sure why, however when I click the Start conversion button on the RVC Realtime GUI, it either freezes or crashes, I have tried various different settings and tried changing the algorithm, but it doesn't seem to work. The input monitor does though...

Pls feel free to ping or dm me

[proper shale]

They couldn't because their feature extraction screwed up

[violet heron]

I don’t think so

[copper pebble]

the 1st and the 3rd

[quaint dock]

why does this happen randomly?

after I make a cover it randomly does this

[proper shale]

translate pls

[quaint dock]

The existing connection was forcibly closed by the remote station

[rare goblet]

Ayo? @quaint dock level 6 !!!

[proper shale]

@sand iris pls

[sand iris]

anything else higher up in the log ?

[quaint dock]

thats in the log, idk why does it close randomly

[sand iris]

Are there any anti virus software that could interfere ?

[night hazel]

If I have an RTX 3050 ti (from a Surface Laptop Studio), what is the optimal batch size, or is it best that I go with Colab?

[proper shale]

How many GBs of VRAM?

[night hazel]

GPU Memory: 12GB
Dedicated GPU Memory: 4GB
Shared GPU Memory: 8GB

So I guess 4GB?

[proper shale]

I'd use batch_size 2 then

[night hazel]

god damn

I used batch size 2 and it was slow af

I guess it made 10 epochs in an hour

I wonder what happens if I do it in batch size 8 since it doesn't give me a "Cuda out of memory" error

[proper shale]

id recommend b_s 6 at that point

i dont really like to set batch size bigger than the dedicated VRAM

[night hazel]

and if I assume correctly the batch size is in GB?

[proper shale]

not realllly

its just that it should be less than ur VRAM

[night hazel]

hm

[proper shale]

or close to it

[night hazel]

alr

[quaint dock]

no, I relaunched it and it doesn't close now

[mint nimbus]

high res time

is it depend on internet?

speed

res is up to 10-20k ms

help pls

[hexed monolith]

how can I create a dataset locally on my pc?

ah, i see the channel nvm

[proper shale]

what gpu do you have

[mint nimbus]

rx 500 serie

amd

downloaded directml version

[wide trench]

im new to this so, what the hell is a rvc 2 and whats the difference between the basic one and it.
Or is it a base to make models or do i actually have to install it or smth to use those models.

[violet heron]

RVC is the program for making and converting models

v2 is the architecture

v2 models work on normal RVC

[proper shale]

Make sure to export every new model you downloaded to onnx

[hexed monolith]

I noticed that when trying to train a model, RVC wasn't able to detect my GPU? How do I fix that

[proper shale]

Do you have an NVIDIA GPU

[hexed monolith]

AMD, but i have the AMD version

[rare goblet]

Ayo? @hexed monolith level 2 !!!

[proper shale]

Well, AMD can't train

[hexed monolith]

its

training right now though

[proper shale]

_ _

it probably isn't

[hexed monolith]

it just used a fallback to CPI

CPU

[hexed monolith]

it is lmao

[proper shale]

itll be painfully slow then

...huh

well it isnt using your gpu so it'll be slow sadly

[hexed monolith]

unfortunate...

[proper shale]

better than nothing tho

[hexed monolith]

better than using Collab too

[proper shale]

nahhh

i think you'd be better off with it

[mint nimbus]

every?

I exported for example one model and there is problem also with him

[proper shale]

yes, cuz if u dont itll lag

[wide trench]

alr so i can just import without worrying abt that. last thing is idk what an epoch is or how to set it 💔

[mint nimbus]

got it

will try, thx

[delicate locust]

Is there still a bot here that does your voice covers?

[violet heron]

Epoch is just training cycles, don’t worry about it

[violet heron]

WordBand

[uneven charm]

Well, I selected the same pth and index as I do for inferencing...

[grim bay]

The index doesn't work

[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

[fair solar]

idk where to ask
so i decided to do 1 ai cover and overall it sounds great but when it comes to chorus part the voice is not well pitched and has glitches(?), can i do anything to fix this, if yes then how?

[violet heron]

Are you using RVMPE?

And are the vocals clear

[fair solar]

i'm not sure what's RVMPE and vocals (if you mean model) are TF2 Scout taken from this server

also if this might help: i'm using rvc gui app for this

[violet heron]

RVC-Gui is outdated

Doesn’t have RVMPE

[torpid loom]

@violet heron whats changed since then

[rare goblet]

Ayo? @torpid loom level 1 !!!

[fair solar]

so i need different tool?

[rare goblet]

Ayo? @fair solar level 1 !!!

[torpid loom]

been outta this for a few months sorry

whats rvmpe too lol

[violet heron]

Basically Everyone uses RVMPE

Which is a new pitch extraction method

Kinda like harvest v2

[torpid loom]

is it better than crepe

mangio crepe

[violet heron]

A lot of cases

Mangio crepe is super sensitive with sound compared to RVMPE

[torpid loom]

im assuming the process to train it locally is similar

its prolly in #1159513888199540817 right

[fair solar]

uhh

so where can i get the tool(?) or whatever i need for that RVMPE?

[brittle wing]

Can someone tell me which site would be advisable (and easy to use) for downloading music in wav or flax to create covers?

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

I checked with VirusTotal the site I used the most, and it showed that it could be malware or spyware with the downloads 😦

(Y2download)

[torpid cave]

i use Yout all the time

it comes out clean and ive never had bad malware experiences with it

it even fills out the metadata for you automatically

also, is there any way to get the elevenlabs other cloned ai voices or make a cloned AI voice without having to pay because that subscription plan is nuts
feel free to direct me to another channel if this was the wrong one to post to

[violet heron]

Google colab

-colab RVC Disconnected I think there’s a guide in #1159513888199540817

[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]

https://y2down.cc/es/youtube-wav.html

Don't download music from YouTube if you don't have to

[torpid cave]

wait would it be easier to make a default voice say something in elevenlabs and cover it with an rvc model

[molten pecan]

Once you upload a song to YouTube, the platform will compress the quality

[violet heron]

Yes

[torpid cave]

ah, i'll take note of that then

[coarse sun]

yooo im using integrated graphics, but it works fine in passthrough

[rare goblet]

Ayo? @coarse sun level 6 !!!

[coarse sun]

can someonbe help me hear the voice changer

because i dont hear it when i try it out, maybe its my settings

[molten pecan]

#🔍│help-w-okada

[coarse sun]

okay thanks

[brittle wing]

Yout?

[brittle wing]

So it's better from other sites like... which others more if it's not Youtube? Deezer or other music sites?

[molten pecan]

Yup I could teach you lots of ways to get lossless quality music

It's pretty easy

I know how to stream it too

Though I can't get into details on the server 😭

[unkempt pasture]

when i try to use it sometimes it gets very choppy any idea why ?

[brittle wing]

Can I send you private message?

Anyways thanks for answer my question guys

[molten pecan]

Yeah!

Go ahead

[brittle wing]

Okay ü

[molten pecan]

Ah nvm I gtg

Sorrt

Sorry

[violet heron]

Covers or real time voice changer

[unkempt pasture]

real time vocie changer

[brittle wing]

Well no problem

[vapid gust]

why taking so long?>

[violet heron]

Did you duplicate the space

And hoe long is the song

[unkempt pasture]

if i post my settings would u b able to let me kno if i needa change anything ?

[vapid gust]

Ye

[violet heron]

Lower chunk

Turn up extra

[vapid gust]

Song is like 3 minutes

[violet heron]

Change f0 to RVMPE

[violet heron]

I think I know why

Audio file has spaces

[vapid gust]

Wdym audio file has spaces

Ofc its gonna have spaces in the

Song

[violet heron]

The name

[vapid gust]

oh

[rare goblet]

Ayo? @vapid gust level 14 !!!

[vapid gust]

Shi

Why would that matter though

[violet heron]

Reload the page and try with that

[violet heron]

Trust me

[vapid gust]

Okok

One sec

[unkempt pasture]

its still cutting off at the end of my sentences

[rare goblet]

Ayo? @unkempt pasture level 1 !!!

[violet heron]

Change in/out and tune

#🔍│help-w-okada

[unkempt pasture]

ok thank you bro appreciate u

[coarse sun]

yo whats the google version

[vapid gust]

Just changed it to no spaces

I’m at 400 seconds

🗿👍🏽

Is it because it is 40mb input file

[violet heron]

Should be same as a mp3

[vapid gust]

Huh

[thick summit]

could anybody help me? it gives me an error and i dont know if its related to the GPU.

ive been trying everything this whole week and theres always an error since the google collabs stopped working...

[vapid gust]

It still don’t work @violet heron

I changed to mp3

Removed spaces

I’m at 400 seconds

[violet heron]

Wait till 550

If it goes past that tell me

[vapid gust]

Alr

[vapid gust]

It errored out

Now it doesn’t even convert

It just shows error

[violet heron]

Try duplicating it again but on an alt account

[vapid gust]

Brooooo

[tropic spoke]

I've been looking at trying AI covers. Does anyone know of some safe local option?

[vapid gust]

@violet heron alt account and still 400+ seconds

I think something wrong with ilaria

@proven hill ilaria no work

[violet heron]

I don’t know what to tell you

Try again tomorrow with a different audio

[brittle wing]

anyone know how to fix the choppy audio

[supple lion]

increase buffer size

[feral sphinx]

I keep getting name Errors can anybody Help me out

[grim bay]

What is the easiest song for the AI to cover in your opinion?

[chilly vessel]

can someone walke me through what to do if I have a model downloaded (pth and index files) and just those downloaded from someone and i want to use my vocals and convert them using the model

[rare goblet]

Ayo? @chilly vessel level 2 !!!

[proud elbow]

put yourmodel.pth file in \weights and added_*.index file in \logs\yourmodel

[chilly vessel]

where are those files

[brittle phoenix]

does anyone have the download for rvc

[chilly vessel]

theres a lot on youtube

[chilly vessel]

can sum1 help me

[brittle phoenix]

you got a yt link?

[chilly vessel]

yh

https://www.youtube.com/watch?v=Q8du7n0vgfU

this should be good

[violet heron]

https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/RVC0813Nvidia.7z

Extract everything and run the bat file

[night bluff]

Please help me, I need to uninstall rvc because I ran out of space to reinstall it

[violet heron]

Just delete it

[feral sphinx]

How can i fix the NameError: name 'experiment_name' is not defined, ?

[rare goblet]

Ayo? @feral sphinx level 3 !!!

[feral sphinx]

Hello?

[long root]

how do i connect voice changer to discord

[violet heron]

Virtual Audio Cable

Set it as your output on Okada

And as your input on discord

[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 silo]

I did it all correct with mine and the discord audio test worked but I’m a vc the audio was just my voice

[short hull]

after I've created my model using the disconnected google colab method, how do i use the model?

[rare goblet]

Ayo? @short hull level 1 !!!

[rare goblet]

Ayo? @brittle wing level 2 !!!

[chilly vessel]

or do u press ctrl c and then type y and enter to close the go webbat file

[proven hill]

you have to duplicate the space

not use alt accounts

[azure marsh]

Documentation

All-In-One English documentation

https://docs.aihub.wtf/

English FAQ

https://docs.aihub.wtf/faq

Credits: AI HUB FRANCE

More guides

How to Make AI Covers using Ilaria RVC

Link: Documentation

Create a model with RVC disconnected (colab)

Link: Google Docs

Credits: Nick088, Angetyde and Kanav

[pale hemlock]

I'm using the CoverGen colab. Is there a way to get it to covert the voices of background vocals?

[zealous shuttle]

is this the best settings for inference? i need reference

[spice rain]

Does anyone know how to upload voices to huggingface?

[finite galleon]

Covergen inference is bad

It leads too a low quality output when compared to something like applio

[zealous shuttle]

you always make cover there?

[rare goblet]

Ayo? @zealous shuttle level 7 !!!

[last patio]

Can someone help me fix this? I can make covers and TTS, but I get this error when trying to create a model.

[W C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\socket.cpp:601] [c10d] The client socket has failed to connect to []:22468 (system error: 10049).
RuntimeError: GlooDeviceFactory::makeUVDevice(): interface or hostname can't be empty

[rare goblet]

Ayo? @last patio level 1 !!!

[tough crane]

I have done the training. These are the graphs that I have provided (stopped at 800 epochs only instead of 1000 cuz colab)

What do you think is the perfect spot to stop training?

[proper shale]

mel is still going down, whilst kl is stagnating but honestly i think that's ok

d might be beginning to stagnate, so I'd use the 33-35k part as of rn - but it is still going, I'd say you should keep training, but if it starts to overtrain right after this, use the 33-35k line

[shrewd spear]

If I press on the "Train-feature-index" on the local version of Mangio-RVC-Web nothing happens, and I cant create a index file. How can I fix this?

[tough crane]

OK thanks!

[proper shale]

np

[proper shale]

have you preprocessed yet

[finite galleon]

Not really but you can do a test using same model, same audio, same setting on aicovergen and on applio

Applio audio will be sharper

[crude jewel]

i have tried using the voice changer and it works on the app but doesnt work on discord any tips?

[silver bobcat]

how to fix cpu run time does change standard time help?

[fervent tusk]

Im trying to run rvc disconnected and when running the first cell i get this error. Any clue how to overcome it?

[mossy fox]

okay so I have a dataset were the speaker does character voices (deep voice, nerd voice, high pitch) would that mess up the training or would it make it better?

[proper shale]

You must run all the dependencies steps at once.

[mossy fox]

its primarly the same voice but there are a few clips

[proper shale]

Disconnect and delete runtime (a button in the Runtime category), then minimize the Dependencies category and just run all the steps at once with the play button.

[fervent tusk]

Thank you. that worked. Does this look right? Im concerned that the % is at zero

[proper shale]

It's right

[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

[solar jacinth]

Let's gooo

[zealous shuttle]

i only had a potato laptop, and there isn't applio colab

[astral jungle]

Yo

[potent root]

whats the name of that software that sorts a datafile by speaker

[shrewd spear]

if this is step 2a yes

[mint nimbus]

How to deletye voices?

remove

leave empty slot

[glacial cargo]

yo the thing always crashe after i press convert

any tips?

[proper shale]

how did it go?

like whats the output info

[brittle wing]

https://github.com/Mangio621/Mangio-RVC-Fork/releases/tag/v23.7.0
https://developer.microsoft.com/en-us/windows/downloads/windows-sdk/
https://visualstudio.microsoft.com/visual-cpp-build-tools/
https://github.com/PowerShell/PowerShell
https://www.python.org/downloads/release/python-3100/
https://vb-audio.com/Cable/ (Required for realtime RVC if you want to route audio so people can hear it)

pip install -r requirements.txt

pip install -r requirements-win-for-realtime_vc_gui.txt

IMPORTANT:
When installing python, made sure to add it to PATH.

How to work it:
After you install everything, you wanna shift + right click the folder with the mangio folder with the requirements.txt IN IT, not a sub folder.
After Shift + Right Clicking, open in powershell and paste the pip commands separately after they complete themselves.
After all is done, run the go-web.bat, to get everything in order, then once the site opens, close it and launch go-realtime-gui.bat.
Refer to screenshot

Using inputs that don't corresponding drivers will not work, like MME and Windows Direct Sound

If there is audio crackling, a simple fix is
go into task manager
go to details
sort by name
find audiodg.exe
set the priority to high
set the affinity to 1 cpu core

how to install rvc

since the tutorial is vague and shit

[glacial cargo]

yo can i maybe get some help?

[copper pebble]

wt r u using?

[glacial cargo]

rvc gui

[copper pebble]

ah

pretty old

[glacial cargo]

what should i use?

[copper pebble]

did u try ilaria rvc

u can also find a guide for it

[glacial cargo]

got a link?

[rare goblet]

Ayo? @glacial cargo level 1 !!!

[glacial cargo]

do i just use the website?

[copper pebble]

https://discord.com/channels/1159260121998827560/1167844377046040769 ilaria rvc (huggin face space)

https://discord.com/channels/1159260121998827560/1175615544037212200 AiCoverGenNOwebUI (colab)

[glacial cargo]

ah

ty

[copper pebble]

duplicate the space is reccomended for ilaria's one

[glacial cargo]

@copper pebble is it supposed to take this long?

[copper pebble]

it takes like 200 to 400s

[glacial cargo]

ah ok

[copper pebble]

it depends on ur audio length

[glacial cargo]

alright

[finite galleon]

Rip

I am working on applio colab infer no ui so

[glacial cargo]

works

[shy zealot]

**Method 1 🔎: ** The search accuracy is 100% 💕

Joseph Stalin

Upload the <t:1698068750>
Server uploaded: AI HUB
Channel uploaded: AI HUB Post
Profile Owner: asasasasasasasasa

Author

E

Discord User

asasasasasasasasa

Internal ID

17751

Language

Unknown 🇺🇳

Type

RVC V2

Size

70 MB

Extractor

N/A

Epoch

250

Steps

0

Powered by ModelAI.net, Made with Love ❤️

**Method 1 🔎: ** The search accuracy is 100% 💕

eggman

Upload the <t:1686222883>
Server uploaded: AI HUB (RIP)
Channel uploaded: AI HUB (RIP) Post
Profile Owner: _thej

Author

the real ultima weapon

Discord User

_thej

Internal ID

235

Language

EN 🇺🇸

Type

RVC

Size

0 MB

Extractor

null

Epoch

200

Steps

0

Powered by ModelAI.net, Made with Love ❤️

**Method 1 🔎: ** The search accuracy is 100% 💕

eggman

Upload the <t:1686222883>
Server uploaded: AI HUB (RIP)
Channel uploaded: AI HUB (RIP) Post
Profile Owner: _thej

Author

the real ultima weapon

Discord User

_thej

Internal ID

235

Language

EN 🇺🇸

Type

RVC

Size

0 MB

Extractor

null

Epoch

200

Steps

0

Powered by ModelAI.net, Made with Love ❤️

[zealous shuttle]

oohh cool

[brittle wing]

oh yeah

[rare goblet]

Ayo? @brittle wing level 1 !!!

[brittle wing]

rvc can run on 4 gb of vram right

realtime

[zealous shuttle]

maybe....if you had a patient

[brittle wing]

rvc

i got a laptop on the side I wanna use as the RVC Realtime so that way it takes the load instead of my main pc

that way demanding games won't stutter it

[spiral osprey]

my queue is infinite pls help-me

[copper pebble]

seems ok

whatcu using??

[violet heron]

Did you duplicate the space

[cobalt oyster]

does anyone have a image which contains which contains appropriate epochs for 10 min 50 min etc

[cobalt oyster]

thanks

[molten pecan]

How accurate is that?

[violet heron]

Not very

[molten pecan]

Is there an epoch guide somewhere?

[glad zealot]

dont thnk so

most of the time tho 300 epochs is a good safe spot to sound decent

[molten pecan]

Interesting

Thank you

[dense drift]

i just trained my model how can i download it now?

should i run this now?

[molten pecan]

#📑│making-models

Try asking on the channel for making/training models

[dense drift]

can someone pls help

[proper shale]

go on the colab file explorer and download the .pth in weights and the index on logs/ur models name

[dense drift]

where?

@proper shale

[surreal bridge]

applio installer doesnt work for me, i press install and it just doesnt start anything, nothing shows in the text area

it just stays here not doing anything, ive tried waiting 10+ mins with no indication of it doing anything

[proper shale]

oh i think its saved on ur drive

check it

[dense drift]

where?

im confussed

[proper shale]

just don't use applio, mangio is better

[proper shale]

on google drive?

[surreal bridge]

i want some of the features applio has tho

[proper shale]

u can try the manual install then

[tame mica]

applio steals features from other forks

[proper shale]

💀

[surreal bridge]

ive tried it before and it never works for some reason

[proper shale]

¯_(ツ)_/¯

really don't know then

[surreal bridge]

installer worked on a diff install of windows i had but it suddenly is broken on a new one

[proper shale]

the whole Applio install is kinda screwed up

[dense drift]

what should i do with this now?

[proper shale]

ooh i think you should train index

[dense drift]

how?

[proper shale]

surely there must be a cell for this

in the colab notebook you're using

[rare goblet]

Ayo? @dense drift level 3 !!!

[dense drift]

this?

or this

[surreal bridge]

anyways, when I try and install mangio every single time without fail ive had to go into it, edit things, reinstall and uninstall dependencies and more because things dont work, the main thing being that it always installs the cpu version of torch when I want it to install the cuda version

any ideas why

[proper shale]

no

there should be a dedicated step for it surely

-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

[dense drift]

WHAT SHOULD I DO NOW OMG

omg

[solar jacinth]

No , it's for start the GUI

Now you have .pth of the model

[proper shale]

ok you probably trained index already

ru using easyGUI? @dense drift

[dense drift]

im using this: https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=FFfC9x239kC1

[proper shale]

yeah that's easyGUI

yeah you've already trained index

[dense drift]

ok

how can i use my model now?

[proper shale]

wait is that all you have on the folder

well whatever, just zip the latest pth (not d/g, they're training files) and the added index

[dense drift]

which one is the index

should i only add the modelname.pth and the index to the file

but which one is the index?

[proper shale]

added_IVF(numbers) and thenbur models name

[dense drift]

i cant find added_ivf

@proper shale

[tame mica]

train index then

[dense drift]

How?

[tame mica]

the notebook..??

[surreal bridge]

applio has no issues with its installers 100%

[dense drift]

It got disconnected

Does that mean that i have to train again?

[tame mica]

no you don't i think

just preprocess feature extraction blablabla again till train index

[brittle wing]

hello?

anyone here?

[worldly knoll]

Hi

Found him

[brittle wing]

Hi

[dense drift]

im using this: https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=FFfC9x239kC1

so how?

[brittle wing]

Hi

[dense drift]

Where can i find feature extracion

[crimson basalt]

Hi.

[brittle wing]

How are you doing today?

[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

[brittle wing]

Hi

[dense drift]

i just trained a model using thi website: https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=3KyMRbK49g__

but i didnt got index

how can i train index?

i clicked on this but it didnt helped

[violet heron]

Check the files on colab

Not drive

[dense drift]

which files?

[violet heron]

Logs folder

the added index file

[violet heron]

Find your way to the logs folder

[dense drift]

which log folder?

where should i click?

@violet heron

[rare goblet]

Ayo? @dense drift level 4 !!!

[violet heron]

Logs folder

The folder is called logs

There should be a .index file that has added in the file name

[dense drift]

but there is no folder called log

[violet heron]

Download that one

[violet heron]

It’s called logs

I forgot where it is

[dense drift]

huh

what should i do now?

[violet heron]

Download the index file that has added from that folder

Along with the pth

[dense drift]

is it that:

[proper shale]

yes

[violet heron]

Yep

[dense drift]

oh

thanks for helping

how can i make ai covers now

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing

this website doesnt work anymore

[proper shale]

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

this

[dense drift]

ok thanks

[cinder fractal]

Yo"

How do we like

Use the models once we've downloaded them

[proper shale]

just use a fork or a colab

[proper shale]

like this one

you just need the link of a model and the link of a song to make an ai cover

[cinder fractal]

Oh

[rare goblet]

Ayo? @cinder fractal level 1 !!!

[cinder fractal]

But we do it online?

[proper shale]

If you have a powerful enough PC, you can do it locally

[cinder fractal]

Wait but like

How do we get the link to a model

from like

#🔍│find-models

[cinder fractal]

?

[cinder fractal]

How do we do that??

I'm very new to Github

[proper shale]

There's a documentation for local stuff if you are planning to use it lcoally

[cinder fractal]

No I mean online

I alr forked it

[proper shale]

ooh

#1159513888199540817 can help too

[cinder fractal]

Okay but can you not help here?

[proper shale]

I mean, I can, but a guide would be much better, I'm not really the greatest at explaining stuff haha

[stuck flower]

Heyy guys! Sometimes when I click "train index file" it simply doesn't do anything. I am following the steps, preprocessing etc etc. It is totally at random. Has this ever happened to you?

[proper shale]

Using locally?

[stuck flower]

Yes!

[proper shale]

So, what does Feature Extraction has as its output?

I'm thinking you might've gotten a "no-feature-todo"

[stuck flower]

['infer/modules/train/extract/extract_f0_rmvpe.py', '1', '0', '0', '/xxxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps', 'True']
todo-f0-55
f0ing,now-0,all-55,-/xxxxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps/1_16k_wavs/0_0.wav
f0ing,now-11,all-55,-/xxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps/1_16k_wavs/1_4.wav
f0ing,now-22,all-55,-/xxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps/1_16k_wavs/3_1.wav
f0ing,now-33,all-55,-/xxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps/1_16k_wavs/4_3.wav
f0ing,now-44,all-55,-/xxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps/1_16k_wavs/6_4.wav
['infer/modules/train/extract_feature_print.py', 'cuda:0', '1', '0', '0', '/xxxxxxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps', 'v2']
/xxxx/Retrieval-based-Voice-Conversion-WebUI/logs/chimps
load model(s) from assets/hubert/hubert_base.pt

[rare goblet]

Ayo? @stuck flower level 3 !!!

[proper shale]

hmmmm

i think feature extraction didnt happen correctly?

how many cpu processes are you using

[proper yarrow]

any one can help me im trying to find whats the easyest way to train an ai model right now🫶🏾

[stuck flower]

hmmmm a lot

lemme tune it down

[proper shale]

keep at 2

always

[proper shale]

#1159513888199540817, rvc disconnected

[stuck flower]

I do have a LOT of cpus, shouldn't be an issue. But kept it at 2 and still didnt change the output

strange

I am running from within a docker container as well

[proper shale]

huh

how many audios are in your dataset

[stuck flower]

not that much, maybe 5 min

actually

way less

1 min

should there be more/

?

[proper shale]

no not the length, i mean how many audios

how many files

[stuck flower]

6 files

[proper shale]

oh

Wait, is the pitch guidance on?

[stuck flower]

yes!

[proper shale]

hmmmmmm

[stuck flower]

Just tried with it turned off... no difference

[proper shale]

so weird

did it ever say "all-feature-done"?

[stuck flower]

not really

I'll try to stick to another machine...... but a bummer

[proper shale]

that is super weird

[stuck flower]

yesterday I trained a model for 10 hours just to realize I had no index file

[proper shale]

oooh that happens

[stuck flower]

And just to confirm, if I do have my pth file after all of that training, I cant infer w/o the index right?

[proper shale]

yeah

but the index is good to have, simulates accents

[upbeat talon]

hello everyone I read the Create your AI modeland have a question Where to get Applio?

[stuck flower]

I found a silent bug in feature extraction, an import error:

ImportError: /usr/local/lib/python3.10/dist-packages/fused_layer_norm_cuda.cpython-310-x86_64-linux-gnu.so: undefined symbol: _ZN3c106detail14torchCheckFailEPKcS2_jRKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE

probably related to Fairseq if it also happens to anyone else!

[desert linden]

Hey, i need help

I tried getting help the other day but im still stuck

I have mac and rvc wont run on my mac

Ive tried every solution possible

[upbeat talon]

hi who can help https://discord.com/channels/1159260121998827560/1176971524377428079

[violet heron]

Which file did you download

[proper shale]

Mangio fork

They can't run the run.sh thing, despite everything we tried

[desert linden]

Ive tried mangio

And now ive tried w-okada

They both r givin me issues

W-okada tells me i cant open the program

Because it says macOs cant identify if the program is sabotage program

Idk how to force it to open

[sand iris]

colab ?

[velvet robin]

I am in desperate need of help finding an easy and effective way to use Text-to-Speech with voice models. RVC is a good choice, but I want to learn the easy way. If there's anyone out there to help me, DM me and I will await your instruction.

I will still use RVC, but I want to look for an effective and easy version of RVC.

Why can't anyone help me in this godforsaken server...

[desert linden]

I seems to have managed to open w-okenda but the program itself is just blank? Does that mean its broken?

[violet heron]

Which file?

[gloomy plaza]

Running with the system Python.
Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases.
Press any key to continue . . . i keep getting that but i have python installed

[velvet robin]

Hmm... This better be good. I'm sorry if I sounded angry.

What's PATH?

Whoops! Sorry about that, I thought you were talking to me.

[gloomy plaza]

what @velvet robin said, what is PATH?

no, i don't think so

i did and now it says "Running with the system Python.
python.exe: can't open file 'C:\Users\NAME\Desktop\rvcgui.py': [Errno 2] No such file or directory
Press any key to continue . . ."

[left totem]

so I have an issue where my response latency just keeps rising

and RVC won't detect my gpu

I am using amd

[gloomy plaza]

i have no idea what that means, can you dumb it down for me?

[sand iris]

for training *

Inference works

[left totem]

anything I can do?

I wanna use the voice changer

just got a new pc

[sand iris]

Wait which software are you trying to run ?

[gloomy plaza]

oh, i can see the .py file, should i reinstall rvc?

[rare goblet]

Ayo? @gloomy plaza level 1 !!!

[left totem]

Voice Changer Client Demo

[sand iris]

#🔍│help-w-okada

There is a guide in pinned messages there

you need to have the right version

there's two different releases for windows, one for Nvidia, one for everything else

[left totem]

I downloaded the amd version

[rare goblet]

Ayo? @left totem level 1 !!!

[sand iris]

you sure ?

[left totem]

pretty sure

since

I checked

and followed the guide on github

[sand iris]

screenshot it

pls

[left totem]

oh I thought I had no image perms

this the program

lemme show which one I downloaded

[sand iris]

That's the nvidia version

https://cdn.discordapp.com/emojis/790214199544512573.webp?size=48&name=kek&quality=lossless

[left totem]

this one

[sand iris]

Yeah you got the wrong version on the link after

no matter which one you click right now, you'll get the full list after

[left totem]

cuz it says this

[left totem]

oh I see

that makes sense

so I get this?

[sand iris]

yes

[left totem]

alr

makes sense

ty

I'll see how it goes

[gloomy plaza]

i have no idea how to install stuff off of huggingface

i would like a model of Jack Walten from the walten files. but i have no idea on this stuff

[velvet robin]

How do I get the requirements listed in requirements.txt?

Uhh... Where do I type that in?

What command do I have to type in with the path?

Nevermind. It's working.

[marsh grove]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English (main)

https://docs.aihub.wtf/

​

Translation by country

🇧🇷 Brasil (PT-BR)

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

🇪🇸 Spain

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

🇫🇷 France

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

🇩🇪 Germany

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

🇯🇵 Japan

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

🇰🇷 Korea

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

🇵🇱 Poland

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

🇷🇺 Russia

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

🇮🇹 Italy

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

[gleaming ledge]

someone knows a solution for this error ImportError: cannot import name 'I18nAuto' from 'i18n' (unknown location) ?

[left totem]

@sand iris res just keeps rising and rising

idk why

[sand iris]

can you screenshot fully again pls

[velvet robin]

Dang, the process is taking to long. The process is going fast, but the problem is that there's so much stuff. Can you belive it?

[left totem]

also I can't pick any of these except two

[sand iris]

"export to onnx"

[left totem]

what about it

[vague sapphire]

does anyone know how to use tortiose tts on collab?

[sand iris]

then reupload the model just like you did, with the .onnx file instead of the .pth one, and don't supply an .index file

[left totem]

oh okay

so I can't use .pth files?

res still keeps infinitely rising

[vague sapphire]

anyone know how to train a model for tts?

[feral sphinx]

Hello i'm am using RVC colab disconnect, but receiving Name Errors and Cells Errors

[desert linden]

Guys, ive downloaded w-okada and it opens the program but the program itself is just blank when it open? And it keeps processing in terminal

But it doesnt do anything else

[brittle wing]

I'd just uninstall and reinstall again

[wraith steppe]

hello

[weary pond]

Hai, I can't use any output/input device from my first PC while using my second PC for hosting RVC, I followed @craggy bramble guide, but i can't figure out how to set the devices up.

[viscid moss]

The same way you train models for RVC.

[vague sapphire]

uhh

[viscid moss]

Disconnect the runtime and start again

[vague sapphire]

is it using hte disconnect google collab link to make it?

[viscid moss]

or try the EasyGUI colab instead

[vague sapphire]

in English: https://docs.google.com/document/d/1XuxQYiqEhYrdYeCZRRLrmV_ciMKo0bV-jTCGHu_-5Cc/edit?usp=drivesdk
this one?

[rare goblet]

Ayo? @vague sapphire level 1 !!!

[craggy bramble]

you should have the settings on client for Okada RVC for input and output

[viscid moss]

Yes, with that you can make the model

[vague sapphire]

okay. can i do it locally if i have amd?

[viscid moss]

No, u need NVIDIA GPU

https://tenor.com/view/gpu-kawaii-rtx-2080-gif-18883392

[vague sapphire]

damn. collab works right with free tier?

[viscid moss]

yep

6 hours per day

for free

[vague sapphire]

samples need to be around 10 secs? couldit be longer?

[viscid moss]

Tesla T4 GPU

[viscid moss]

all the dataset splitted in 10 seconds audio files, ya

[vague sapphire]

woah my questiouns are being answered so qquick lol feels great

https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI
does this work the same as the goole collab thinkn

[viscid moss]

ya, but locally

[velvet robin]

@heady tendon I got good news and bad news.
Good News: The process has been finished
Bad News: 2 errors occurred, here's the error list.
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for fairseq
Building wheel for antlr4-python3-runtime (pyproject.toml) ... done
Created wheel for antlr4-python3-runtime: filename=antlr4_python3_runtime-4.8-py3-none-any.whl size=141217 sha256=ae53055b50502c4b17e0963914fda1e0368821f19b117f82c2e11d3eaaed2280
Stored in directory: c:\users\tristan\appdata\local\pip\cache\wheels\a7\20\bd\e1477d664f22d99989fd28ee1a43d6633dddb5cb9e801350d5
Building wheel for ffmpy (pyproject.toml) ... done
Created wheel for ffmpy: filename=ffmpy-0.3.1-py3-none-any.whl size=5601 sha256=99599867698e81f4e64a4313f0ced13863b3c37a00f4d2b3422b303b038c0784
Stored in directory: c:\users\tristan\appdata\local\pip\cache\wheels\01\a6\d1\1c0828c304a4283b2c1639a09ad86f83d7c487ef34c6b4a1bf
Successfully built antlr4-python3-runtime ffmpy
Failed to build fairseq
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects
Like what the Command Prompt said, the error on the top is not quite an issue but I have doubts with the one in the bottom.
Let me know if that error is an issue.

[rare goblet]

Ayo? @velvet robin level 7 !!!

[vague sapphire]

i see

[viscid moss]

u can make an inference but no train

no compatible GPU

[vague sapphire]

what is an infernce?

[viscid moss]

change the voice into the model's voice

like a cover

[vague sapphire]

ohhh

this ai stuff is hard to grasp

[weary pond]

on the browser window i can only see the inputs and outputs of the second PC

[rare goblet]

Ayo? @weary pond level 1 !!!

[viscid moss]

what the heck

[velvet robin]

I know. It's a lot.

[vague sapphire]

when training there is a setting cache all training checkpoints.
this is for saving the epochs?

[viscid moss]

I can see a fairseq problem, maybe the version or something like that

[craggy bramble]

Is this what you have
PC1: Client Okada RVC browser
PC2: Running Okada RVC

[weary pond]

exactly like that

[craggy bramble]

Then your client should be able to access the input and outputs of PC1

[viscid moss]

For rvcDisconnected u can save only when the training is done

[weary pond]

yeah i thought so too...but i can't see them

[viscid moss]

For the EasyGUI idk

[craggy bramble]

take a screenshot

[velvet robin]

Does anyone know how to open this RVC version?
https://github.com/litagin02/rvc-tts-webui

[weary pond]

this is on PC 1, as an Input i need a device called AT2020 and it's not there, as any of my devices on the first PC

[vague sapphire]

https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=DZDKirCM0F9g

this collab link also does the same thing?

[craggy bramble]

you're on server not client

[weary pond]

gimmi sec

[craggy bramble]

it's showing you the input and output of PC2 because you're on server and not client settings like I said

[viscid moss]

have u tried this?

[viscid moss]

yep, train and also inference

[vague sapphire]

tensorboard something i can use to determine quality of themodel?

[viscid moss]

yep

but.....

[velvet robin]

No, I haven't tried that. I'll see if it's useful.

[weary pond]

Where's the swich to server side?

[viscid moss]

For that colab it seems you need an ngrock account, for the other one you don't

[vague sapphire]

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

is this the other one?
or disconnect?

[viscid moss]

-colab

[azure marsh]

Suggestions for @vague sapphire

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

[craggy bramble]

Go to PC1 on your Okada RVC browser look for the settings CLIENT and SERVER. CHECK CLIENT NOT SERVER

the settings are above input and output

[viscid moss]

@craggy bramble To add something to Automaze I have to talk to Ray right?

[craggy bramble]

talk to any of the admins

[viscid moss]

ty

[sand iris]

screenshot cmd log

[weary pond]

those?

[craggy bramble]

yes choose client

[weary pond]

[sand iris]

firefox / brave moment

[vague sapphire]

is learning ai or machine learning topics beneficial for this kind of stuff?

[rare goblet]

Ayo? @vague sapphire level 2 !!!

[sand iris]

as an end user ? ehh, not really

[vague sapphire]

hmm

do you guys teach how to make deepfakes?

[craggy bramble]

make sure to run on a chrome browser

[weary pond]

oh Opera don't works?...