✨│ai-help

[trail iris]

https://tenor.com/view/vocaloid-dance-kasane-teto-teto-territory-cute-gif-17052544

[violet heron]

i need hep

[jolly geyser]

hep

[brittle wing]

help

[thin sparrow]

Just use wordband bot

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

[mental lava]

Bruh

[thin siren]

what is wordband do?

[tulip fog]

It creates music with words

[plucky locust]

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

[thin siren]

https://tenor.com/view/facepalm-really-stressed-mad-angry-gif-16109475

[old wigeon]

why is mine so delayed and echoey

[ember ermine]

hello

i would like to try and recreate snake's voice from MGS but for voice effects and not necessarily speaking (like the grunts/hmmms he does in combat). is it viable to use RVC to train on voice data that doesn't have any words? instead just voice effects

[glad zealot]

Yes but why

[ember ermine]

for more freedom in what can be done with the voice

[rare goblet]

@ember ermine has reached level 1.

[ember ermine]

ty anyways. also, does the crepe-hop setting matter when training with the rmvpe algorithm?

the colab i'm using has that parameter even when rmvpe is selected

[plucky locust]

melanie martinez is not a registered model! Use /list to list all models, or use /register to register a new model.

[tawdry sage]

/list

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

[carmine hearth]

Is there Kaggle for RVC training?

RVC Disconnected exist so

[rare goblet]

@carmine hearth has reached level 1.

[lament eagle]

Does anyone know how formant shifting works

[rare goblet]

@lament eagle has reached level 2.

[lament eagle]

or a doct that explains it or smth

[red kayak]

We haven't really played around with it yet since it's still a work in porgress

U may use the pre existing settings

[signal oracle]

do you know what i'm doing bad?

[elder vortex]

Try:

/content/drive/MyDrive/dataset/Dinseyl.zip

[signal oracle]

ok!

[honest pond]

how i get my rols?Yesterday they were given to me but today they were taken away.

[honest pond]

.zip

and not /dataset if is desconected

[honest pond]

what colab are you using?

[rare goblet]

@honest pond has reached level 1.

[signal oracle]

what should I enter?

[honest pond]

what colab do you use? do you have it in content or drive?

[signal oracle]

and should I use a packed dataset (zip) orfolder with wav ?

[rare goblet]

@signal oracle has reached level 1.

[honest pond]

prob is a orfolder

but yes

[honest pond]

try this

because i didnt think you move the dataset to content

open drive in the folder in you left

[signal oracle]

ok!

let me check

[signal oracle]

with colab do you prefer to use?

[honest pond]

i not in google drive

but yes, you have a folder into a folder

is not /MyDrive/Dataset

[signal oracle]

oh!

[honest pond]

is /MyDrive/idk what meens that/dataset/Disnet.zip

@signal oracle open the folders in your left

[signal oracle]

so what can i write?

[honest pond]

rigth clic in your zip if you find

and copy path

[signal oracle]

ok

whu colab cant see zip?

its shows only empty folder

[honest pond]

?

[honest pond]

you cant see the folders in that zip

[signal oracle]

yup

[rare goblet]

@signal oracle has reached level 2.

[honest pond]

but copy the path (rigrh clic and copy path) of the zip and put there

[honest pond]

on dataset folder

[signal oracle]

i know

i trained lots of model

but i dont have colab pro and i must open another colab

[honest pond]

idk

i see meny people using rvc desconected

[signal oracle]

i dont think its good colab..

[honest pond]

rvc desconected=

?

[signal oracle]

do you have link?

[slate wren]

Which is the best Program to do ai covers? Applio or something different?

[signal oracle]

for training or render?

[honest pond]

nah i try but i dont find

[honest pond]

Appilo v3

[signal oracle]

oh...

[signal oracle]

it's for colab pro?

[slate wren]

Dont know the difference, pretty new

not trying to train some voices

just making songs for fun

[signal oracle]

i use rvc-gui local

for making songs

and mixing in FL studio

[slate wren]

this?

[honest pond]

is local

[rare goblet]

@honest pond has reached level 2.

[honest pond]

you have colab pro?

[signal oracle]

oh! can i have link?

[slate wren]

Can you send me the link to Appilo v3, when i tried downloading it yesterday it didnt work

[signal oracle]

YEP!

[honest pond]

yes wheit

[signal oracle]

no but i think to buy it!

[slate wren]

What does colab do

what is it good for

[rare goblet]

@slate wren has reached level 1.

[signal oracle]

https://colab.research.google.com/

[honest pond]

https://github.com/IAHispano/Applio-Installer/releases

[slate wren]

ohh i see

[honest pond]

better download de v4

[signal oracle]

if u have bad computer you can use it! I use it becouse i can train lots of models haha

[slate wren]

Oh thats very nice, but gladly i have a very good computer

[signal oracle]

me too haha

colab has become paid, so I will move to local and I will only be able to train one model at a time

[slate wren]

Idk why but the AI Covers i make sounds very strong that you know its AI even tho i use the same model as someone else

and for him it sounds really real

[signal oracle]

can i hear it?

you can send me it on priv

[slate wren]

sure

[red sleet]

Can someone explain to me what “epoch” is and it’s significance

[slate wren]

its been like this for the past 20mins

is it normal?

[honest pond]

you do clock in install and you have git?

[slate wren]

I clicked install and yes i have git

Wait i update git and try again

[honest pond]

yeh

[slate wren]

ah now it works

[honest pond]

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS

rvc desconeted

[signal oracle]

oh thx very much!

[rare goblet]

@slate wren has reached level 2.

[slate wren]

now its stuck here

[slate wren]

@honest pond idk what to do

after I click install, it just says

Cloning into 'F:\Applio'...
Cloning process completed with exit code 0
Finished cloning repository

and nothing else happens, doesnt show ny other text, what should i do?

[proper shale]

I think it should be ok

[cold raptor]

someone is using my model

[sharp apex]

is there any notebook to make covers with AI? besides colab?

[slate wren]

Ofc, we love bausi

does someone know where to put the index file

in what folder

[cold raptor]

trust me download applio

[cold raptor]

create folder TheBausffsBySztef

and add index here

in logs

like this

[slate wren]

ahh alr

ty very much

also thanks for the epic model

[sharp apex]

there a link?

[rare goblet]

@sharp apex has reached level 2.

@violet cipher has reached level 1.

[cold raptor]

https://github.com/IAHispano/Applio-Installer/releases

[sharp apex]

aight thanks

[junior zenith]

Hello guys! https://github.com/elcolex777/rmvpe-ai/blob/main/Retrieval_based_Voice_Conversion_WebUI_v2_kaggle.ipynb for train model.

[slate wren]

i have the instrumental in

but how to i put the ai voice in

to combine both of them

[exotic ember]

Where are the models generated by Applio saved?

[slate wren]

Applio-RVC-Fork\assets\audios\audio-outputs

[woven atlas]

Is rvc in colab is available in smart phone

?

[honest pond]

steel need help¿

[slate wren]

no

but i need help with this

How do i combine instrumental and the voice

[honest pond]

idk if i remember you need to go in interference

upload the audio there

and then you will be able to choose audio.

[slate wren]

how

I dont understand

whats the interface

[honest pond]

go up

the firt page

home page

[slate wren]

ok

[honest pond]

there where you can put you audio

[slate wren]

i fixed it

[rare goblet]

@slate wren has reached level 3.

[honest pond]

the big rectangle

[slate wren]

thanks

[brittle wing]

using rvc gui

[viral sentinel]

hey prolly a dumb question but im using rvc gui and im trying to start making better AI voice modeled music, anyways just wondering what "feature retrieval rate" would do and what the difference is i couldnt tell much from trying myself

[junior zenith]

I recommend install Applio Rvc fork local version or Replay.

[thorn spear]

how do i train my own models

[errant dagger]

why are we talking about replay again all of a sudden?

like it was desolate before

[past turret]

is apllio local now suggested over mangio rvc local?

ive used mangio rvc no problem, but does applio have new features?

[winter seal]

wait whats @shy zealot

[sand iris]

Use whatever works lol

Stuff can be unstable so just stick to what works, applio doesn't really offer anything groundbreaking at the moment

Just a more cozy UI and different colors, but it's the same features

[lethal sand]

hey guys can i have colab link to use (idk if am on the right channel)

[sand iris]

Yeah that's the right place

[sand iris]

For making models ?

[lethal sand]

no for making covers also maybe models

[sand iris]

Oh, there aren't any colabs for that at the moment

only for making models

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
I think that is the right one but I could be wrong

This notebook is not designed to be used for inference. Colab should not be used for RVC inference in general, use local inference instead. (RVC can inference just fine on any dime-a-dozen CPU from the past 7 years, you seriously don't need Colab.)

[lethal sand]

ohh okay

ig kits.ai or voicemy.ai is my only options

[plain bolt]

I have a question, is it possible to use rvc on a phone? Or is it possible to do this ai song cover on a phone?

I don’t got a pc which is why I’m asking

[sand iris]

Nah you can do conversion just fine without a beefy computer

just takes a while

That's what the colab notebook author explains

[sand iris]

Unfortunately no

[lethal sand]

k but i was talking abt making ai covers

[plain bolt]

F and the only options I can find on phone require spending money for those memberships

[sand iris]

Yeah, he explains why there aren't any cover-making colabs (it's because any computer can convert audio just fine)

[lethal sand]

k

can i have download

[sand iris]

Oh fuck we don't have the guides and downloads here yet

[brittle wing]

Hello guys, when I try to start easy gui locally, this is what it says: "Is not recognized as an international or external command, operable program or batch file"

[burnt zealot]

any1 trains locally and know what batch size i should use for a rtx 3060?

[rare goblet]

@burnt zealot has reached level 1.

[thin sparrow]

Use the wordband bot

[plucky locust]

Community Models

goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more

Check out all of the public community models at https://www.word.band/voice_models

You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

To see your private models and register new ones, please start a DM with me!

[lethal sand]

k

[sand iris]

Oh wait yeah the bot exists

I'm dense

Sorry y'all

[fossil basin]

Hey can someone help my with how to set up the voice changer for discord if anyone has some time this week to walk me through it one on one

Just dm

[lethal sand]

My character isnt there tho

[brittle wing]

how can i make my own model

[visual turtle]

#1159290666019471421

[thin sparrow]

You can register a new one

With a huggingface url

Using /register new

[brittle wing]

Hello guys, when I try to start easy gui locally, this is what it says: "Is not recognized as an international or external command, operable program or batch file"

[visual turtle]

Do you want to train a model or do you want to do a cover?

[brittle wing]

I mean both

I downloaded the Train thing so I can do both I think

[brittle wing]

So is there any way how to fix it?

[rare goblet]

@brittle wing has reached level 1.

[fossil basin]

So can anyone help me

[loud hazel]

For ?

[fossil basin]

Hey can someone help my with how to set up the voice changer for discord if anyone has some time this week to walk me through it one on one with

[loud hazel]

@fossil basin go here https://discord.com/channels/1159260121998827560/1159290161683767298

[fossil basin]

O ok thank I will ask there

[errant dagger]

yall I fuckin forgot my models number

[shy zealot]

Sonic

Upload the <t:1685892959>
Server uploaded: AI HUB
Channel uploaded: AI HUB Post

Author

bush did 911

Discord User

_thej

Internal ID

176

Language

ES 🇪🇸

Type

RVC V2

Size

155 MB

Extractor

N/A

Epoch

200

Steps

0

Powered by ModelAI.net, Made with Love ❤️

[errant dagger]

the bot is wrong

[rare goblet]

@errant dagger has reached level 2.

[errant dagger]

I set it to english

[mossy pasture]

i am suprised you are still here lol

[loud hazel]

@mossy pasture where i Can go....;)

[grave sail]

any downloads for mac?

[mossy pasture]

applio?

[rare goblet]

@mossy pasture has reached level 1.

[woven atlas]

Is colab is going to be able to inference again?

[normal karma]

Th

Anks

[wind mantle]

Its based on your vram / virtual memory size. If you have 12gb you can safely use 20 batch

[burnt zealot]

i used 16 and it took longer though

[wind mantle]

Taking longer is ok. Lower usually takes longer

It will straight up crash if it's too high

If it took longer than a lower value not sure

[burnt zealot]

yea it did take longer than lower value

[burnt zealot]

do yk what i put for this tho

[wind mantle]

It means python isn't recognized. Usually python is installed on system path or within the root of the application somewhere

[wind mantle]

You can use your max cpu, but it can slow down your pc. I've used 24 threads and didn't have issues.

The more threads, the less it strains your cores

Generally you use as many threads as you have cores

If you use more it risks slow downs or crashes but usually is fine

[burnt zealot]

id rather use 8 bc i have 8

ty

[woven atlas]

?

[fleet viper]

Using Colab No UI...

[snow jolt]

yes

[royal folio]

Some collaboration to do interference :((

Collab*

[brittle wing]

And can I fix that?

[brittle jay]

whats this mean

i just tried submitting a model in #1159514067187277865

nvm got my answer

but i have another question, how long is the bot gonna be down for?

[prime hemlock]

a bit

[flat zealot]

how do i install applio

its been on this for 10 minutes

nvm

[rare goblet]

@flat zealot has reached level 1.

[flat zealot]

fixed

i just didnt run the installer

[valid elbow]

is there a way to make rvc gui work with AMD gpu

[rare goblet]

@valid elbow has reached level 1.

[fading summit]

i tried following the guide and watch youtube videos but i still dont know how to use apollo or mangio-crepe

to install or use i

[flat zealot]

how long does this usually take

i've been waiting for 20 minutes

nvm

[short aspen]

Do The Wieght.gg Voces Still Work And If So How Do I Get Them To Work?

[radiant laurel]

How can I make a model I tried Jarods Journey how to and it still doesnt work please send help

[shy monolith]

is it doing somethign

i dont know if its done or not

you know this channel does not have a lot of people helping

99% of it is people asking questions, 1% is answers

[radiant laurel]

yeah....

[brittle jay]

is there a way to make covers with colab?

i already can make models with rvc disconnected, but is there a colab like that?

anybody know?

?

[daring verge]

-colab

[azure marsh]

Google Colabs

RVC

AI-Cover-Gen-No-UI

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

Note: This colab is for inference only.

RVC v2 Disconnected

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS

Note: This colab is for training only.

RVC colabs translated to other languages

• PT-BR: -colabs_br

Audio Separation/Isolation

MDX Net Colab

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

UVR arch colab 5.0.2/c1.04

https://colab.research.google.com/drive/16Q44VBJiIrXOgTINztVDVeb0XKhLKHwl

KaraFan colab

https://colab.research.google.com/github/Captain-FLAM/KaraFan/blob/master/KaraFan.ipynb

[brittle jay]

Now which of these is for making ai covers

Nvm

But I do need help with the ai cover Gen colab

[shy monolith]

ok for some reason i cant send images

[rare goblet]

@shy monolith has reached level 1.

[shy monolith]

but i downloaded a model using the applio resources tab

and now i dont know what to do

its not in the model inference tab from what i can see

[sudden wasp]

whats the best pitch extraction for a dataset with only speaking voices?

[final wave]

Guys, I don't remember how to post a model on huggin face

help me

😭

[nocturne mural]

Verify that it has been correctly downloaded and unzipped by going to the logs/weights or logs/name folder of the model.

[shy monolith]

ye i did that

i was running the python file wrong

but now im having another issue

AttributeError: 'NoneType' object has no attribute 'tobytes'

when i try to convert

[nocturne mural]

send the complete error

[shy monolith]

Attempting to load assets\audios\cc.wav....
2023-10-06 04:14:23 | WARNING | lib.infer.modules.vc.modules | Traceback (most recent call last):
  File "D:\Applio-RVC-Fork\Applio-RVC-Fork\lib\infer\modules\vc\modules.py", line 375, in vc_single
    audio_opt = self.pipeline.pipeline(
                ^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\Applio-RVC-Fork\Applio-RVC-Fork\lib\infer\modules\vc\pipeline.py", line 636, in pipeline
    and os.path.exists(file_index)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "<frozen genericpath>", line 19, in exists
TypeError: stat: path should be string, bytes, os.PathLike or integer, not list

Traceback (most recent call last):
  File "C:\Python311\Lib\site-packages\gradio\queueing.py", line 388, in call_prediction
    output = await route_utils.call_process_api(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gradio\route_utils.py", line 219, in call_process_api
    output = await app.get_blocks().process_api(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gradio\blocks.py", line 1440, in process_api
    data = self.postprocess_data(fn_index, result["prediction"], state)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gradio\blocks.py", line 1341, in postprocess_data
    prediction_value = block.postprocess(prediction_value)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gradio\components\audio.py", line 349, in postprocess
    file_path = self.audio_to_temp_file(
                ^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Python311\Lib\site-packages\gradio\components\base.py", line 325, in audio_to_temp_file
    temp_dir = Path(self.DEFAULT_TEMP_DIR) / self.hash_bytes(data.tobytes())
                                                             ^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'tobytes'```

[nocturne mural]

ah , press f5 to gradio and select again your index and pth

[brittle jay]

where is the AI generated cover at when this cell gets done generating?

[shy monolith]

hold on can you go to dms

cos i cant send images here

[nocturne mural]

ok

[slow summit]

Been looking after the Adam Smasher model. has that been taken down completely? Hugging Face says "Repository not found"

[brittle jay]

can someone please help me out.

where is "SONG_INPUT"

[plush tree]

hlp i can't install applio since my graphic card was intel i3

[brittle jay]

can someone get on this channel and help everybody out?

[plush tree]

I have Intel i3 graphics card and the repository doesn't work for me

[brittle jay]

omg i found it, it was right under my nose, stupid me

[spiral osprey]

please help me the app is giving me 504 gateway timeout

Can anyone allow uploading images here?

[wide ingot]

It seems gradio is down. Can't ping to it as well.

https://status.gradio.app/

[spiral osprey]

Do you know how soon he should be back?

[rare goblet]

@spiral osprey has reached level 1.

[wide ingot]

The "Share Links" section is down for over seven hours, according to the site. However, it's hard to say when it will be back up and running.

[spiral osprey]

some voices left the roll and the levels were reset, what was changed on the server?

[wide ingot]

The CLI is bugged, so I can't use.

[spiral osprey]

I was rendering voice through the disconnected RVC. for me it hadn't fallen

[spiral osprey]

what is CLI?

[wide ingot]

Command-line interface. You type something, and it'll do some stuff.

[spiral osprey]

I think I got kicked off the server.

[spiral osprey]

understand

[wide ingot]

The whole server got taken down yesterday. This is the new one.

[spiral osprey]

understand

I had noticed because I had left the server for no reason and the voice I wanted a while ago was no longer there. I thought I had been expelled because I had been banned for breaking the rules lol what a shock.

[wide ingot]

Reasonable.

TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
 1. Downgrade the protobuf package to 3.20.x or lower.
 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
make: *** [Makefile:56: run-ui] Error 1

If you see this error while starting Applio/Mangio on Paperspace, append PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python before it like this:
PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python make run-ui

Also applies to TensorBoard.

[rare goblet]

@wide ingot has reached level 3.

[spiral osprey]

For me this doesn't appear, it only shows 504 gateway time-out

[wide ingot]

Same here. Gradio share links as a whole is down. (0.000%)

The share API still works if one knows how to code in Python. I don't, as I mainly code in C.

[spiral osprey]

Does disconnected RVC still work?

[wide ingot]

That Colab should still work.

[spiral osprey]

If so, does it still work?

[rare goblet]

@spiral osprey has reached level 2.

[hexed hull]

where is latest RVC Easy GUI colab file?

[quartz heart]

cant i send photos?

the rvc is just loading and loading

and not working D;

[hexed hull]

same issue, may be error from gradio.live

[uncut nebula]

Same

[glad zealot]

I don't think any trainer works with gui atm

[rare goblet]

@glad zealot has reached level 2.

[glad zealot]

The colab ones

[quartz heart]

do you know when it wil work?

[uncut nebula]

Not really

[real wind]

Can someone make a full YouTube video on how to train a model for free ?

[rare goblet]

@real wind has reached level 1.

[serene musk]

https://youtu.be/MfjKeE_s7mE

[quartz heart]

google collab is dead bro

[bleak bluff]

hey guys i want to train a voice where can i do that?

[quartz heart]

that video not gonna help him

[snow pasture]

I've used CoverGen for the first time ever and the whole points issue that pops up is basically a troll since you don't need points for getting anything there.

Detailed explanation :

1. It literally covers songs
2. You can get clean vocals, backing and instruments extracted
3. It applies your model to main vocals by default
4. You can download everything and do the editing yourself.

[serene musk]

Yeah local is the way

[rare goblet]

@serene musk has reached level 2.

[serene musk]

Haven't used a colab since may

[bleak bluff]

some one @ here ?

[rare goblet]

@bleak bluff has reached level 1.

[verbal ridge]

Can someone help me get Mangio-RVC working locally?

[snow pasture]

CoverGen really needs a zip file download since it's annoying having to download everything one by one before it times out forcing a redo of the whole thing to get the rest...

[quartz heart]

do you think by any chance you can train a model for me pls?

[verbal ridge]

Trying to open the GUI gives me this message:

Traceback (most recent call last):
  File "S:\Mangio-RVC\EasierGUI.py", line 28, in <module>
    from my_utils import load_audio, CSVutil
ImportError: cannot import name 'CSVutil' from 'my_utils' (S:\Mangio-RVC\my_utils.py)```

[azure vine]

Gradio.live not opening today, is any one facing issues ?

[tulip fog]

You running it through the command prompt or just click on it?

If one doesn't work try the other

That's like my go to python fix

[verbal ridge]

I'm running it by opening the run_easiergui.bat file

[tulip fog]

Oh that's what you're supposed to do. But try it in the command prompt see what happens

[verbal ridge]

Alright, gimmie a minute

[tulip fog]

It's because the my_utils.py from Mangio-RVC-Fork is missing CSVutils.

[tulip fog]

It's because the my_utils.py from Mangio-RVC-Fork is missing CSVutils. You need to get the my_utils.py from kalomaze/Mangio-Kalo-Tweaks.

The direct link to the updated my_utils file is below.
https://github.com/kalomaze/Mangio-Kalo-Tweaks/blob/main/my_utils.py

click on the "download raw file" on the top right corner to download it. Aftewards, copy that file to your Mangio-RVC-Fork folder and overwrite the previous my_utils.py file. It should work after.

There you go courtsey of GitHub and a user named mellow-tnk who deserves all the credit for figuring this out back in July

[dawn sedge]

whats the difference between rvc v1 and v2?

[tulip fog]

Rvc 1 is old and most models are being made for 2 I believe. I also heard someone say RVC v2 uses more resources so if you don't have a good PC rvc v2 may run slightly worse but it will result in a better product in the end

[lean owl]

does anyone have a guide for rvc disconnected colab? im refering to the colab link in the info channel

its confusing to use and i dont understand it

[verbal ridge]

It's now opened great but all the operations run endlessly. Can't download a model or even refresh the model list.

https://i.imgur.com/N68vsEj.png

[tulip fog]

Shit I was getting that same issue before

[verbal ridge]

What solved it? Or did it just disappear?

[lean owl]

can i have help here?

[azure vine]

Gradio live not working?

[tulip fog]

I deleted it and haven't used it since

So probably not the answer you want to hear, maybe try copying all the files into the folder again?

Something could've gotten corrupted.

Although I doubt it

I'm gonna look into in the meantime see if I can find something

[verbal ridge]

I just installed it less than an hour ago 🫤

Could something have actually happened in that time?

Maybe I should try a reboot

Of my PC

[tulip fog]

Hey you mind showing what the command console looks like? It might show an error message or give a clue

Yea try that first, but I'd love to see the command prompt and see if there's any error messages or anything out of sorts

[lean owl]

how do i use rvc disconnected colab?

[tulip fog]

No hurry tho I'm not going anywhere, just make sure to tag me cuz I'm very forgetful

[tulip fog]

Honestly I got no idea bud, I've never tried it.

[verbal ridge]

S:\Mangio-RVC>S:\Mangio-RVC\runtime\python.exe EasierGUI.py --pycmd S:\Mangio-RVC\runtime\python.exe --port 7897

-------------------------------
RVC v2 Easy GUI (Local Edition)
-------------------------------
Not running on Google Colab, skipping download.
Found GPU NVIDIA GeForce GTX 1060 6GB , force to fp32
Pretrained weights not downloaded. Disabling training tab.```

[lean owl]

yeah its kinda confusing

[verbal ridge]

Isn't everything labeled?

[tulip fog]

You have a better PC than me so it's definitely not a hardware issue

Glad we can rule that out at least

[verbal ridge]

I thought the 1060 was becoming below average these days

Brb

[tulip fog]

Okay so it is but my brain always switches the 10** series and the 16** series

But I have a 1650 with less vram than you so idk

But let's keep trying, right now we're gonna try my personal strategy of throwing shit at the wall and see if it sticks

[verbal ridge]

Ok no administrator mode this time and I'm giving it all the firewall perms it wants

[woven atlas]

Traceback (most recent call last):

File "/content/AICoverGen/src/main.py", line 11, in <module>

import gradio as gr

ModuleNotFoundError: No module named 'gradio'

1

[rare goblet]

@woven atlas has reached level 1.

[woven atlas]

Why I get this error

?

[glad zealot]

Cause you don't have gradio installed

[woven atlas]

How I'm doing that on colab

[tulip fog]

Ive heard some people say to just turn off the antivirus when you're installing it

[verbal ridge]

Didn't really install it, I unzipped it

[tulip fog]

Same difference

I'm not a fan of turning antivirus off

But people say it helps sometimes and rvc is weird. It's always some random things that fix it

[verbal ridge]

I don't have an antivirus aside from Windows Security, and I don't see an off button there anywhere

[tulip fog]

Real time protection, first menu in windows defender

[glad zealot]

You're probably using a banned Collab then

[tulip fog]

It's the one I have so I can see it in my head without even having my PC open

Windows defender rocks!

But not when you're doing stuff like this

[verbal ridge]

Defender Firewall?

[tulip fog]

No just real time protection

I'm gonna do some more searching

[wide ingot]

Gradio is back

[verbal ridge]

This is all I see: https://i.imgur.com/BG8hJVh.png

[glad zealot]

Since you're doing covergen use this
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb

[tulip fog]

Windows security that's it

The one with the white shield for the logo

[verbal ridge]

Oh I was just there

Oh found it!

[tulip fog]

Yay! You didn't even need me lol

Just make sure to turn it back on once you're done. I don't want you getting a virus

[verbal ridge]

Ok gonna retry from scratch

(Assuming my internet will let me)

[tulip fog]

Just unzip the file again

You should need to redownload everything

Shouldn't***

[red gyro]

i have a 1650ti laptop 🙃

[verbal ridge]

I'm gonna redownload because I watched a video about how Microsoft can see inside and scan the contents of .zip files. People were getting their quarantined virus samples deleted by Microsoft in their cloud storage.

[tulip fog]

Me too lol

Minus the ti

Yea that happens to me all the time since I'm a hacker

Like just yesterday it ate one of my Trojans right out of my folder after I explicitly told it not too

And in doing so the trojan took the defender out with it and I had to fix both

Because it has a antivirus kill switch

The whole thing was a big mess so good call on your part

[glad zealot]

Did they turn off real time protection?

[tulip fog]

I saw it on GitHub

And I trust GitHub with my life

Seriously if some of the stuff I got off there fails I'm going to big boy prions

So I trust this will hopefully fix the issue

[verbal ridge]

It's not like I was getting the files from Peter, and I trust Kalomaze + AIHub admins at this point

[tulip fog]

Yea and i'm not gonna mess with someone who has my favorite month in their name!

[rare goblet]

@tulip fog has reached level 11.

[verbal ridge]

Based

In 2 languages, no less :)

[compact spoke]

how do you run local on mac

[verbal ridge]

Actually Japanese has no word for July, and you can't spell it cleanly with Katakana either, so 七日 literally translates as "7th month".

[verbal ridge]

#📜│rules

[tulip fog]

You speak Japanese?

That's pretty cool

[verbal ridge]

Oh no no, I was just following the server trend Lusbert started

I learned a bit though

Man... after all of this, it still can download/refresh models

Total virus protection is still off

[tulip fog]

Dumb question but you have the required models in the correct folders right?

Turn the virus protection back on then , sorry that it didn't work .

At this point I'm dumbfounded

Hey can you link me to guide you used. The more info I have the more likely I am to have one of my clasisc breakthroughs

[verbal ridge]

I'm in the "download models" tab so I figured I didn't need to manually place models anywhere. But even if I did, I would assume it would still be able to allow me to refresh models to check if I had any.

Every button just buffers endlessly

[tulip fog]

Yea you gotta put the index and pth in the correct folders

[verbal ridge]

Then why is everything else not responding, and why is there a a download button?

[tulip fog]

Yea I don't know that's why I gotta see the guide lol

It'll let me know what I'm supposed to tell you to do next

But if you haven't used a guide then a link to the GitHub page

[verbal ridge]

I'm using the EasyGUI, like in the images

I'm gonna call it a night. Can you DM me instead if you find anything?

[tulip fog]

Sure thing, I'll try downloading it on my PC and see what happens. I'm at my best when I'm hands on

[woven atlas]

Thank man it's really helped me

[brittle jay]

what did you do? make an AI cover?

[gilded surge]

how to download this?

[proper shale]

#1159618293594861579 message

[odd remnant]

How fast is just the basic paperspace membership compared to colab

[proper shale]

The $8 one is actually faster, I believe

Depending on the GPU you get ofc

[sand iris]

Do y'all know if there's anything random about model training ?

[rare goblet]

@sand iris has reached level 6.

[sand iris]

If I give it the exact same data, will I get the exact same model back ?

Assuming the same pretrains, f0 extractor and stuff is used

[proper shale]

no

Because it's a GAN

There's always a bit of randomness when training

[sand iris]

Is there a way to set the seed used then ?

I'm being told mangio & applio have humming issues while training and I'm highly skeptical about it and just wanted to do some tests (I don't think the training code was fundamentally altered)

[proper shale]

Only if you do the training code manually

[rare goblet]

@proper shale has reached level 11.

[proper shale]

It's weird lol

But I think on the command it has a seed for it

[sand iris]

Have we ever heard about humming issues to begin with ? Never saw it mentioned

[proper shale]

I think we have, probably an issue with the pretrained models.

I have, at least.

Lol

[wide ingot]

The humming sound is quite noticeable on some but not on others. It sounds like the European electrical hum of sorts.

[woven atlas]

?

[rare goblet]

@brittle jay has reached level 3.

[ember ermine]

does rvc support training on non-word phonemes? if i train a model on hurt/grunt/yell sounds from a videogame character then recreate the same noises during inference as an input would i actually get something useful or can it only do words from languages?

[proper shale]

you would get similar results yeah

[ember ermine]

oh cool! thank you

[rare goblet]

@ember ermine has reached level 1.

[ember ermine]

should i seperate the two? like have one model for the dude talking and another for his combat sfx each with their respective datasets and use them seperately or can i just keep the dataset mixed with talking and noises and have one model?

[proper shale]

You should mix both

[ember ermine]

very well. thank you

[brittle wing]

Does anyone know how to combine voice models?

[proper shale]

You can't

[graceful obsidian]

ckpt fusion

[rare goblet]

@graceful obsidian has reached level 1.

[proper shale]

_ _

CODENAME?????

[graceful obsidian]

yea

still haven't recovered but hell

[proper shale]

omg we missed you

[graceful obsidian]

how so?

[visual turtle]

简单易懂

[graceful obsidian]

well, that's just to represent how it works

[proper shale]

i want this 💀😭

[graceful obsidian]

but it's been in rvc for a long while now

or so I think at least, my saturn fork / poet is still outdated so idk

[proper shale]

oh wow

[graceful obsidian]

should update it as well, curious of that whole rmvpe+

[rare goblet]

@graceful obsidian has reached level 2.

[proper shale]

rvmpe+ is just rvmpe but you can change the minimum/maximum pitch

i guess thats cool

[graceful obsidian]

hmmm, i suspect it to be different from standard octave playing

[proper shale]

it uses hz and not notes

but the original experimental version did use notes

[graceful obsidian]

hmmm, that's nice, more fine-tuning capabilities then

[proper shale]

yeah

[brittle jay]

hey, Im making a Goku voice model, but the dataset has grunts in it, will that affect how my model sounds when its done training?

[proper shale]

Not really

It'll make grunts sound less weird tho

So definitely keep em

[brittle jay]

even the top-of-lung screams?

like when he's roaring to go super saiyan or somethign?

[proper shale]

yea

[brittle jay]

okay, thanks

[woven atlas]

Yes and train

[vague sedge]

I never trained a model 😆

Did any of you guy use audio webui

[daring verge]

-colav

-colab

[azure marsh]

Google Colabs

RVC

AI-Cover-Gen-No-UI

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

Note: This colab is for inference only.

RVC v2 Disconnected

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS

Note: This colab is for training only.

RVC colabs translated to other languages

• PT-BR: -colabs_br

Audio Separation/Isolation

MDX Net Colab

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

UVR arch colab 5.0.2/c1.04

https://colab.research.google.com/drive/16Q44VBJiIrXOgTINztVDVeb0XKhLKHwl

KaraFan colab

https://colab.research.google.com/github/Captain-FLAM/KaraFan/blob/master/KaraFan.ipynb

[dense kite]

what do i have to do to get a model working

[sonic garden]

model maker?

[exotic ember]

Completed the training of the model, where to find a model with a certain epoch, how to make an index to it?

[robust needle]

Inside RVC directory, logs/ contain the index files, and weights/ contain the .pth model files

[rare goblet]

@robust needle has reached level 1.

[robust needle]

You'll have to click on 'train feature index' in RVC gui

[exotic ember]

What if I didn't do it?

[robust needle]

you'll go back, add the exact name of the model in the training tab, the path for dataset and then click train feature index

It's not an issue, I also forgot to train it a few times, just start up RVC again, and follow the above instructions. As long as you have the model's folder in logs/ you can train or re-train feature index any time

[frail cave]

sup, anyone tell me where make ai covers and models actually? same rvc v2

[exotic ember]

I use Applio

[rare goblet]

@exotic ember has reached level 1.

[brittle jay]

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS heres the link to train models, but idk what to tell you for making covers, but use the public applio link, (I use the applio colab to make like 3 covers before it disconnects, when it disconnects, simply delete and restart the runtime, heres the applio colab link: https://colab.research.google.com/drive/157pUQep6txJOYModYFqvz_5OJajeh7Ii

[exotic ember]

What is the difference between added and trained index?

[proper shale]

#📜│rules and #1159513888199540817

[proper shale]

Added is the one that works

that's all you need to know

[brittle jay]

yeah download added

[exotic ember]

Thanks

[frail cave]

why "no works" in fork applio??

[ember ermine]

is the crepe hop length parameter relevant in training if i'm using rmvpe as the pitch extraction algorithm?

it's at 64 by default and i'm wondering if it has any effect if i change it and if i should

[brittle jay]

i think its cause of the short runtime before it disconnects? but if your quick enough, you'll be able to do covers of 3 songs before it does disconnect

when it does, simply delete and restart the runtime and wait for the public gradio link to appear again

it still works, but the runtime is just really, really short

but still fast enough to make like 3 song covers

..that was the first time I've actually guided someone in this server lol

[brittle wing]

start preprocess
['trainset_preprocess_pipeline_print.py', '/notebooks/Mangio-RVC-Fork/datasets/', '48000', '4', '/notebooks/Mangio-RVC-Fork/logs/jematria', 'False']
/notebooks/Mangio-RVC-Fork/datasets//.ipynb_checkpoints->Traceback (most recent call last):
File "/notebooks/Mangio-RVC-Fork/my_utils.py", line 134, in load_audio
ffmpeg.input(file, threads=0)
File "/usr/local/lib/python3.9/dist-packages/ffmpeg/_run.py", line 325, in run
raise Error('ffmpeg', out, err)
ffmpeg._run.Error: ffmpeg error (see stderr output for detail)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/notebooks/Mangio-RVC-Fork/trainset_preprocess_pipeline_print.py", line 84, in pipeline
audio = load_audio(path, self.sr, DoFormant, Quefrency, Timbre)
File "/notebooks/Mangio-RVC-Fork/my_utils.py", line 141, in load_audio
raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: ffmpeg error (see stderr output for detail)

[proper shale]

should be .wav, not zip on dataset

[brittle wing]

it's not .zip

[proper shale]

ig it didnt extract

[brittle wing]

that's the thing

[proper shale]

well what file type is this

[brittle wing]

mp3

i tried with wav too

same thing

[proper shale]

huh

how long are those?

[brittle wing]

12m

a single 12m file

[proper shale]

are you sure dataset path is correct?

[livid linden]

I'm trying to make an AI cover using No UI Google colab, but at the end, it keeps telling me to go Collab pro or to restart runtime.
It doesn't work, does anyone know why its happening?

[brittle wing]

fixed it by creating a subfolder in datasets

[proper shale]

That one has been restricted, I believe - check #📜│rules for other Colabs

[brittle wing]

should i use 32 batch size if the gpu has 16gb

[proper shale]

no

god no

[brittle wing]

paperspace, not colab

[proper shale]

still

[brittle wing]

what should i use then

16?

[proper shale]

thats way too high

go for like 4, 8

i suggest trying out multiple batch sizes and seeing which one has the steadiest slope, in the beginning of the training

[brittle wing]

error here, on terminal it's going tho

[rare goblet]

@brittle wing has reached level 6.

[brittle wing]

is that bad

[proper shale]

yeah no its doing it

[brittle wing]

oh wait im dumb

[proper shale]

dw

[brittle wing]

it was still doing feature extraction and i was tryna train 💀

[proper shale]

_ _

it happens to the best of us

[spare fog]

Do any of the new colabs have tensorboard?

[proper shale]

yeah, rvc disconnected especial

especially*

[spare fog]

Thanks

[ember ermine]

does crepe hop length matter when training with rmvpe?

[proper shale]

nope

[ember ermine]

very well thanks

[frail cave]

anyone know error convert rvc v2 locally?

[rare goblet]

@frail cave has reached level 1.

[frail cave]

i click convert and "error"

[proper shale]

what does cmd say

check cmd

[frail cave]

brasileiro?

[proper shale]

ss

kkkkkk

[frail cave]

consegue vir dm?

mais facil

-rvc

[azure marsh]

RVC Guides (how to make ai cover)

🍏 Applio-RVC-Fork: Guide (EN/ES)

https://docs.google.com/document/d/12YmIQFLFzth1YsZJMGrA0aeTN0Go8DNZVw8LFSf-Hug

Tip: You can also watch this video tutorial made by iroaK if you need further help.

RVC v2 Tutorial for Paperspace (By LollenApe)

https://docs.google.com/document/d/1lIAK4Y0ylash_1M2UTTL_tfA3_mEzP0D2kjX2A3rfSY/edit?usp=sharing

RVC Guides in Other Languages

• Russian FAQ for RVC/Voice changer
• Italian Guides or -it_g
• RVC Guides in PT-BR (-rvc_br)

[proper shale]

blz, é melhor q os gringos fiquem aq msm

[frail cave]

ss

[terse nymph]

AMD applio local won't actually process the voice and most times just errors 😭 idk how to fix it

[hazy orbit]

How can I upload Audio File in it ?

[frail cave]

RVC V2 Local convert error, can anyone help me?

[visual turtle]

try full reinstall magnio fork

[frail cave]

i make this, and nothing

[visual turtle]

pip install gradio --upgrade

[thorn abyss]

I had this error

make sure that there are no non-Latin characters in the path to the file (and in the file itself)

[hazy orbit]

????????????

[frail cave]

whats "non-latin"

[thorn abyss]

cyrillic

or any

only a b c .. n

and spec sym

[spare fog]

When I run the training part of the Lemonfoot colab, I get a Google 403 error box instead of tensorboard. This also happens when I try using the old colab.

[copper pebble]

gys what can use for TTS with models??

[brittle wing]

is this overtraining?

found the overtraining image

[civic laurel]

how do i use rvc

[proper shale]

reduce smoothing

[civic laurel]

wait

how do i open this file

[crimson basalt]

Does mangio-crepe crash others system with applio?

As in training

My system gets a blue screen crash when using mangio-crepe feature extraction

[proper shale]

thats like colab pages, dw

[proper shale]

what GPU do you have

[crimson basalt]

4070ti works fine on magio rvc but crashes on applio

[wet egret]

Blue screen can mean anything

[proper shale]

just try rvmpe tbh

as your feature extractor

works well for good or bad datasets

[crimson basalt]

Yeah that's what I'm doing atm. Just found mangio-crepe gave cleaner results.

[proper shale]

is more noise sensitive tho

i feel like mangio is very restrictive, hence why rvmpe is kinda better

[civic laurel]

i tried to install the file at paperspace but it kept telling me to buy their gradient

[rare goblet]

@civic laurel has reached level 1.

[proper shale]

oh

yeah you have to subscribe to get anything decent from them

[civic laurel]

is there a free one

[proper shale]

nope

[civic laurel]

i mean alt one

[proper shale]

not really

[civic laurel]

ok

[half cove]

hmmmmm

[late raptor]

rvc

[proper shale]

?

[spare fog]

Anyone know how to solve this?

[rare goblet]

@spare fog has reached level 1.

[proper shale]

I actually had the same issue yesterday...

I hope it's fixed once I use it today because it's literally for a commission

@rigid mango Do you know about this issue?

[rigid mango]

It’s a Colab-Chrome bug that notebook authors can’t fix

https://github.com/googlecolab/colabtools/issues/3990

I ended up switching to firefox solely because of this bug, it’s annoying

[terse nymph]

hi, what rvc [[VOICE APPLYING PROGRAM]] can let me upload my input files, not a model, but my isolated vocals from a song

every one i use wants a youtube link, and isolates it horribly

[twilit gulch]

is there a way to train without utilizing my specs (they make me bluescreen 💀)

like just not my GPU/CPU yk?

[honest pond]

how i get my rol helpper again

[vagrant marsh]

use gpus:
C:\Users\PC\Documents\RVCv2v23.7.0\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\python.exe train_nsf_sim_cache_sid_load_pretrain.py -e CarolPingreyTLH -sr 40k -f0 1 -bs 1 -te 600 -se 10 -pg pretrained_v2/f0G40k.pth -pd pretrained_v2/f0D40k.pth -l 1 -c 0 -sw 1 -v v2 -li 26
C:\Users\PC\Documents\RVCv2v23.7.0\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\cuda\__init__.py:651: UserWarning: CUDA initialization: The NVIDIA driver on your system is too old (found version 9010). Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver. (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\c10\cuda\CUDAFunctions.cpp:109.)
  return torch._C._cuda_getDeviceCount() if nvml_count < 0 else nvml_count```

is it a error?

[twilit gulch]

@vagrant marsh Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx
d installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver.

basically the error states (I believe)
The NVIDIA driver on your system is too old (found version 9010)

[vagrant marsh]

okay, I'm gonna update it thanks

[zinc anchor]

Yeah thankfully the error message is straightforward in this case

[twilit gulch]

if im wrong my apologies I just went on reading the message

[vagrant marsh]

NVIDIA is smart. Making an portable driver

to download

[twilit gulch]

#✨│ai-help message
?

[vagrant marsh]

well, my gpu is gt 610

created in 2012

[twilit gulch]

its not a gpu thing

[rare goblet]

@twilit gulch has reached level 1.

[twilit gulch]

it's an update software thing

[vagrant marsh]

oh

okay

thanks

[twilit gulch]

👍

[zinc anchor]

There are options for online training on #📜│rules

[twilit gulch]

👍

https://cdn.discordapp.com/emojis/1135994557042012230.webp?size=48&name=thanks&quality=lossless

[zinc anchor]

Yw

[twilit gulch]

damn, I understand nothing

whats the difference?

do I use the colabs? I assume yes?

[chilly moon]

We are allowed to post speaking samples of our models right? Just no vocals or instrumentals of songs?

[twilit gulch]

backing vocals stacked onto AI vocals should be legally allowed if they're quiet enough, just no instruments.

instruments are to (probably) be distributed in dms

[zinc anchor]

Are you doing training or inference?

[twilit gulch]

I want to make my own AI model (I have my own samples)

[zinc anchor]

I suggest RVC Disconnected

[twilit gulch]

may I ask why such name?

[zinc anchor]

Idk exactly

[twilit gulch]

anyways, thanks for the suggestion

I need to boot my files to a google drive yeS

?

[zinc anchor]

Yeah

[twilit gulch]

will there be a setting to uh

delete the excess files in training

automatically

[zinc anchor]

It doesn't have the backup feature that the older banned Colabs had

[rare goblet]

@zinc anchor has reached level 5.

[twilit gulch]

but theoretically it shouldn't crash or anything with large samples?

I mean

large as in the whole file

[zinc anchor]

You can get disconnected from the runtime but it won't crash usually

[twilit gulch]

if I crash can I uh

resume it by reopening it

or like reinputting some file

how does it work exactly

[zinc anchor]

Sorry but I have to go, ask another helper if necessary

[twilit gulch]

alr

[proper shale]

Ello

So

To restart training you gotta set the same training variables + batch_size and all that stuff

So, same settings...

[twilit gulch]

but

will it have to restart everything

or will it auto resume

progress

[proper shale]

It'll auto resume

[gusty hound]

you know who else will auto resume? MY MOM!

[proper shale]

From the latest checkpoint

[twilit gulch]

sick

[proper shale]

Good for you

[twilit gulch]

ima make drama CD DIO ai voice

I'll message here if I fuck something up 💀

[gusty hound]

you know who else will message here if you fuck something up 💀? MY MOM!

[proper shale]

_ _

[proper shale]

Ok, I'll be here probably so dw

[twilit gulch]

🙏

[rare goblet]

@twilit gulch has reached level 2.

[twilit gulch]

also what does colab utilize

spec wise

[proper shale]

A T4

[twilit gulch]

whats a T4

[proper shale]

It's a GPU

Pretty decent one actually

[twilit gulch]

virtual one or?

oh shit

how many epchos do I do

total_epochs whatever

epochs

[proper shale]

Go for 100

[broken thicket]

whenever i start the vc my res slowly spikes up to like 20000 and basically delays whatever i say by 20 sec

anyone knows the reason

[proper shale]

And when u retrain you can change that

[proper shale]

What GPU do you have?

[vagrant marsh]

It doesn't work

[broken thicket]

geforce 1660 super

[twilit gulch]

how long does that take

[proper shale]

Varies a lot

[twilit gulch]

also whats a batchsize

[proper shale]

Can you send your settings?

[broken thicket]

it used to work fine for the first few days i used it but then suddenly just started delaying everything

[twilit gulch]

u sure u didn't fry it

[proper shale]

Basically the amount of data itll get per GB of vram iirc

[twilit gulch]

what should I make it

leave to default?

[vagrant marsh]

This drivers doesn't work

[proper shale]

F0 Det should be RVMPE or RVMPE_Onnx, and Extra should be lowered

[proper shale]

I'd recommend 10/12 for fast training and not much quality loss

[distant forum]

my voice ai doesn't sound at all like the character. do I have some settings wrong or something

[proper shale]

This is an old driver

[wooden lotus]

Yes

[distant forum]

are you serious right now bro?

[vagrant marsh]

So where I can find a new Driver?

[rare goblet]

@vagrant marsh has reached level 2.

[proper shale]

What GPU do you have, @vagrant marsh

[vagrant marsh]

GT 610

[wooden lotus]

Yes

[proper shale]

oh

yeah it won't do AI stuff then

[twilit gulch]

I don't really want speed, only want best of best quality

[proper shale]

you'd need a 1070 at least

[twilit gulch]

I'm used to having slow speeds since my main hobby is extracting with UVR

[proper shale]

4 it is

[twilit gulch]

I see.

[vagrant marsh]

So, I won't clone voices anymore?

[twilit gulch]

and total_epochs overtraining have any average value so I don't go overtrain it?

[vagrant marsh]

Colab needs computer units

kaggle is banning rvcs

[distant forum]

I have good hardware too

[rare goblet]

@distant forum has reached level 1.

[glad isle]

What apps do I have to install in order to run the ai voice thingy

[proper shale]

Check #📜│rules , these work

[twilit gulch]

is tensor board always supposed to be this error 404 thing

[wooden lotus]

Yes

[twilit gulch]

I reloaded it (I thought it'd maybe fix it)
(it didn't)

[glad isle]

what is this

[vagrant marsh]

I'm gonna check it, thanks

[twilit gulch]

its spewing out code so if there's any error, it will log it there I assume.

[glad isle]

aight

thanks

[twilit gulch]

anyway, for testing the actual model what do I use?

[wooden lotus]

Yes

[twilit gulch]

is it another colab or something

[wooden lotus]

Yes

I use the easy gui

[distant forum]

should this be max?

[wooden lotus]

But the colab seem not to be working

[twilit gulch]

link it?

[wooden lotus]

May try /cover as a alternative

Or the applio

[twilit gulch]

/cover?

I like your words funny man

[plucky locust]

Sorry! You need to register an account to continue making free covers. Please use the /login command.

[wooden lotus]

I am not joking

[twilit gulch]

what the fuck

my /commands aint doing shit

[plucky locust]

Sending you a private login link

[wooden lotus]

skill issue

Just check your discord

[twilit gulch]

I'll just do it on my phone

[plucky locust]

Sending you a private login link

[wooden lotus]

Damn wordband went down

[twilit gulch]

should I be using like 5000 epochs?

[wooden lotus]

Nah

[twilit gulch]

I did 500

[wooden lotus]

Like a few hundreds

[twilit gulch]

but I have a pretty huge model

[wooden lotus]

Yes

[twilit gulch]

like

[distant forum]

does ''ai voice'' app work with these ai voice models?

[twilit gulch]

super huge

[wooden lotus]

I suppose so and not

They may use different methods or the same one

Like scv so-vits

[distant forum]

so what should I use?

[wooden lotus]

Wdym by that?For what purpose?

[plucky locust]

You're in the queue! Your cover will be ready soon. Number of people in front of you: 0 (Status: 🟢)

[twilit gulch]

thats alot of files you know

[wooden lotus]

What

[rare goblet]

@wooden lotus has reached level 5.

[wooden lotus]

Real

[twilit gulch]

right?

[distant forum]

to use ai voice models

[wooden lotus]

Like how many save per epoches you set?

[twilit gulch]

uhhhh

I think I should've done 50

but I did 20

[wooden lotus]

The usual is 5 or 10 idk

[twilit gulch]

tf

fr?

[wooden lotus]

You said 500 epoches right?

[twilit gulch]

yes

[rare goblet]

@twilit gulch has reached level 3.

[twilit gulch]

it'll only take like 20-30 mins

[wooden lotus]

https://cdn.discordapp.com/ephemeral-attachments/1159290139609137264/1159900720976363520/cover.mp4?ex=6532b4b9&is=65203fb9&hm=3c4b4003e8ce721cab9428188725b22256ab71d15cfc57d2a4b171c6e58ec6fa&

[twilit gulch]

4 hours on my pc 💀

[wooden lotus]

The cover works fine

[twilit gulch]

ok but like

my voice guy sounds nothing like a human

[wooden lotus]

Haven't touch any about training

So idk something changed or not

[twilit gulch]

https://youtu.be/OBcJQCMt3Tk?t=2604

(this isn't where I got my samples, only an example of his voice acting)

(too compressed for me)

[wooden lotus]

I won't listen to that

Goodnight

Time to sleep

[twilit gulch]

I timestamped it

:bru:

[wooden lotus]

What

[twilit gulch]

u would've just needed to hear those 5 seconds

[wooden lotus]

Ok what about it

[twilit gulch]

he sounds barely human

don't know if it will work

[wooden lotus]

Can i explain it like you're five?

[twilit gulch]

yes please

[patent onyx]

guys i need no ui rvc

im using one rn and it keeps giving me an error

[wooden lotus]

Ai hear voice
Ai learn how voice sounds
Ai try to make some audio sound like the voice

Like laugh won't work well

[twilit gulch]

ofc

laugh wil lalter it

I tried to remove the majority

bbut it will still like him even if it's not human>

?

[wooden lotus]

/cover

idk wdym by that

[burnt agate]

breathing, laughing and screaming/overly loud voices can cause issues with models from my experience

[patent onyx]

i meant a colab notebook

[rare goblet]

@patent onyx has reached level 1.

[patent onyx]

anyone have a colab i can use

[twilit gulch]

https://drive.google.com/file/d/1XWgpRdZ8hzLAiXEdQKwYqnu4mwV_1e6l/view?usp=drive_link
this is what I'm using, feel free t ocheck it out (it's 450 files)

[wooden lotus]

Goodbye chat

[wooden lotus]

Nah like isn't that dangerous your info can be leaked

[twilit gulch]

yes https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=ZodNcumpg-JM

[burnt agate]

all working colabs are here
#📜│rules message

[twilit gulch]

what

it's just my file

[wooden lotus]

Idk tho

Not a tech guy

[patent onyx]

not for training

for covers

[twilit gulch]

/covers

[patent onyx]

i was trying https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb#scrollTo=B3BqnUoePVGd but it keeps running an index error

[wooden lotus]

What the hell even this

Is this the dataset

[twilit gulch]

That im using? Yes

[final stream]

-colab

[twilit gulch]

Why? Why not

[azure marsh]

Google Colabs

RVC

AI-Cover-Gen-No-UI

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

Note: This colab is for inference only.

RVC v2 Disconnected

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS

Note: This colab is for training only.

RVC colabs translated to other languages

• PT-BR: -colabs_br

Audio Separation/Isolation

MDX Net Colab

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

UVR arch colab 5.0.2/c1.04

https://colab.research.google.com/drive/16Q44VBJiIrXOgTINztVDVeb0XKhLKHwl

KaraFan colab

https://colab.research.google.com/github/Captain-FLAM/KaraFan/blob/master/KaraFan.ipynb

[wooden lotus]

Oh ok

Estimate by length not files

[final stream]

-colabs_br

[azure marsh]

Colabs RVC traduzidos para PT-BR

AI-Cover-Gen-No-UI_para_YouTube

• Para inferência com vídeo do YouTube

AI-Cover-Gen-No-UI_upar_seu_AUDIO

• Para inferência upando seu próprio áudio

Nota: Os dois colabs acima faz a separação da voz do instrumental, aplica a voz modelo e junta novamente com o instrumental

RVC v2 Arrumado para Treino

• Acesse o colab para treino neste link

💡 Dica: Use o comando -rvc_br para guias em Português

[burnt agate]

I wouldnt reccomend posting your dataset like that, someone might yoink it

[twilit gulch]

how can I tell