✨│ai-help

[open ruin]

iam facing this issue anyone can help

[wispy burrow]

RVC models and TTS models are not the same.
In Applio there is a section for TTS.
It first generate a voice with TTS model and then it Infers that with the RVC model.

[surreal bridge]

applio doesnt even work for me

is there some sort of text to speech people would recommend that I can then use as the audio with an rvc model

[prisma kettle]

anyone can fix this problem?

[rugged shuttle]

@prime hemlock how do i make a ai voice model

what do i do

[rancid knoll]

https://docs.google.com/document/d/13ebnzmeEBc6uzYCMt-QVFQk-whVrK4zw8k7_Lw3Bv_A/edit#heading=h.bjzhhhcn3f69 follow this document

[astral perch]

@rancid knoll at what epoch do you usually pause training to check if the model is overtraining

[wispy burrow]

That's probably a W-Okada thing...
#🔍│help-w-okada

[prisma kettle]

thxx

[rare goblet]

@prisma kettle has reached level 2.

[stoic nimbus]

kinda not related but vanguard messing up my audio driver and now my headset echoes, uninstalling riot client and vanguard doesnt fix it. anyone have the same problem as mine?

[prime hemlock]

gotta sell your soul to burger king

[rugged shuttle]

applying for model qc wish me luck https://cdn.discordapp.com/emojis/859048635383808010.webp?size=48&name=salute~3&quality=lossless

[prime hemlock]

https://tenor.com/view/super-mario-headbang-head-banging-rocking-out-devil-horns-gif-26520087

[clear hound]

whats a good ass voice changer

[rare goblet]

@clear hound has reached level 2.

[stoic cloud]

goodluck

[low shard]

what about one for separating 2 different voices singing at the same time? i usually use 6_HP-Karaoke-UVR and put vocals as the input

[rare goblet]

@low shard has reached level 15.

[burnt agate]

I actually have not tried that yet

[low shard]

if you try it keep me updated please

[burnt agate]

alrighty, you have any examples I can test on?

like just a song name or something lol

[low shard]

honestly i dont know,last time i used that was on an undertale fan song one of my subs requested me to do ,i feel a bit embarassed about it but the song was undertale the musical its raining somewhere😭

i dont really remember other songs to test rn as i dont usually do songs with many singers tbh

hope you can test this on some songs and wish u the best for the results

[burnt agate]

no worries will test and see what I can do

[astral perch]

I don't have a noise remover on UVR5

[burnt agate]

did you download the model in the download center?

[astral perch]

What do you mean by download center

[burnt agate]

click on the wrench icon next to **start processing **

[astral perch]

wich model do i need to download

[wispy burrow]

Guess I'll need to update my UVR (5.5)...

[burnt agate]

have a look here
#✨│ai-help message

[burnt agate]

time to get 5.6

[burnt agate]

so Karaoke 2, 5hp and 6hp can split multiple singers but 6hp is the best imo when it comes to quality on this song I just tested

[rancid knoll]

i usually set the epochs to a really high value then check while the training is running

[low shard]

thank you man

[burnt agate]

btw you might wanna make some changes on https://discord.com/channels/1159260121998827560/1163210088618664006

[low shard]

wdym?

OH

DAMN

SHIT

[burnt agate]

rip lol

[low shard]

I POSTED IT 20 MINS AGO WTF HAPPENED

ima fix dis shit

[astral perch]

You can check the graph while the model is training?

[rancid knoll]

yeah you can refresh the tensorboard graph while its running

[astral perch]

I see

The graphs imo are still so confusing lol but I'm happy I learned more about them

[stoic nimbus]

how can i stop rvc from picking up voice from the headphones??

[wispy burrow]

For Realtime Voice Changing you're supposed to go to #🔍│help-w-okada

[stoic nimbus]

oh ok im sorry

[rare goblet]

@stoic nimbus has reached level 1.

[astral perch]

Means it needs more training right?

[stoic cloud]

It should look like it's going down very fast at first, and then it goes down slower and slower

Roughly

Of course there's some noise

[astral perch]

Is that on both on G and Dloss

[wispy burrow]

GLossTotal is the most important one.

[astral perch]

I was talking about this but I still find it very hard to recognise overtraining

Since there are a lot of dips

[wispy burrow]

As long as it generally goes down, it should be good.
Also, make sure you increase "smoothness" for the graphs if you haven't done so.

[astral perch]

So even when it goes up but goes down after doesn't mean it's overtrained right. Training it more will still help

[surreal delta]

why isnt rvc detecting my gpu

[hushed ledge]

So I know that a good gpu is the thing that matters the most for rvc but does the microphone matter at all? Is there any advantage between a good standalone microphone and a cheap headset?

[spiral osprey]

Can anyone clear my doubts? How do I do a cover with 2 people singing, in this case a feat? Anybody know? Or is there a way for me to make a voice model there with 2 people without making a bug?

[naive fossil]

set 0

python counts numbers from 0 not from 1

so 0 doesnt means 0, but 1

[terse nymph]

so im trying to apply a voice to vocals but it keeps giving me this distorted pitch-snappy sound, like really bad autotune

i remember older versions of rvc just distorting it, how do i keep it from doing this in general?? theyre very clear vocals im applying to, so i dont get why its doing this

mostly happens with sliding notes

[rare goblet]

@terse nymph has reached level 3.

[brittle wing]

Best hop length? Default is 64, but how low is too low?

[placid holly]

What New in that version?

just notice new update 5 days ago

[brittle wing]

can anyone tell me how to kill the localhost on MAC? just quiting or reseting terminal doesn't work..

[terse nymph]

seriously, how to fix this??

these are not wobbly notes by any means

i cant see why its so autotune

[terse nymph]

yes that is indeed the vb im using

now how do i fix the autotuning, the model nor song has that in it

i assume its just distortion but due to new applio stuff it just makes it sound like... that

[brittle wing]

Try a different software?

[terse nymph]

bro

im on AMD only applio has support for that so far and it took so long to get it working

[brittle wing]

Try a differnt voice and see if you get the same result?

[rare goblet]

@brittle wing has reached level 2.

[terse nymph]

yup it does

[terse nymph]

ok, i popped the vocal input into Levelator to normalize it, it fixed every single crackle. ✌️

nevermind it didnt fix the end.

[bold ermine]

How can k solve ffmpeg error while training

[brittle wing]

can anyone help me with the applio app? i still dont know how to fix the "'runtime\python.exe' is not recognized as an internal or external command"

i installed python already but still dont know how

[worldly knoll]

Hi

https://discord.com/channels/1159260121998827560/1163388780397469797

[high night]

This is help brotha

Not #1159289738314919936

[loud summit]

One message removed from a suspended account.

[stoic cloud]

hi

[proper shale]

its like 10am UTC ppl are still sleeping

• i don't think intel has support

[rare goblet]

@hushed bison has reached level 3.

[proper shale]

maybe Mangio has support but I'm not sure

[brittle wing]

Idk

[proper shale]

Wait what Intel GPU do you have

[brittle wing]

Try

Btw can anyone help with my applio? Just says runtime\python.exe is not recognized as an internal or external commands

There is python in it

The latest version of it

[proper shale]

Have you added Python to path during the installation?

[brittle wing]

Wdym?

[proper shale]

There's an option for it iirc

[brittle wing]

Which path btw?

[proper shale]

An option named "Add to path" or something similar

[brittle wing]

I'm new to it so you might need to walk me through it

[brittle wing]

Oh

Yeah I think I'll try it later

[proper shale]

Eh I'm not really a local user so I don't know a lot

Just gotta wait till it builds

[brittle wing]

Is there a Google colab version for it?

[proper shale]

#1159949278270193734

[fluid horizon]

is 40-ish minutes of dataset enough for a model? 😄

[high night]

yeah

[rare goblet]

@high night has reached level 17.

[high night]

more than enough

[fluid horizon]

thank you!

[brittle wing]

still doenst work 😔

how to separaet voices in a podcasr

[rare goblet]

@brittle wing has reached level 2.

[oblique venture]

im using rvc v2 on collab, when i load the dataset it gave me this error

FileNotFoundError                         Traceback (most recent call last)
/usr/lib/python3.10/shutil.py in move(src, dst, copy_function)
    815     try:
--> 816         os.rename(src, real_dst)
    817     except OSError:

FileNotFoundError: [Errno 2] No such file or directory: '/content/temp_dataset/sample_Yumare/Yumare' -> '/content/dataset/Yumare'

During handling of the above exception, another exception occurred:

FileNotFoundError                         Traceback (most recent call last)
3 frames
/usr/lib/python3.10/shutil.py in copyfile(src, dst, follow_symlinks)
    252         os.symlink(os.readlink(src), dst)
    253     else:
--> 254         with open(src, 'rb') as fsrc:
    255             try:
    256                 with open(dst, 'wb') as fdst:

FileNotFoundError: [Errno 2] No such file or directory: '/content/temp_dataset/sample_Yumare/Yumare'

how to fix it guys

[brittle wing]

it is possible for RVCv2 to text-to-speech without .npy file?

[fluid latch]

be more specific

[brittle wing]

me?

[fluid latch]

mhm

[brittle wing]

so, I was watching a tutorial on YouTube and I realized some of voice models must requires a .npy file, what I downloaded there were only two types of files, namely .pth and .indext and I doubt that it's possible without .npy file, so can it work even without the .npy file?

[fluid latch]

what tutorial

[brittle wing]

how to separate multiple voices in podcsdt

[brittle wing]

adding a voice models and set it on model_info.json

[fluid latch]

i think thats a so-vits-svc model

you cant use that with rvc

like an rvc model only needs a .pth file and .index file

if any has .npy or .json files its not an rvc model

[brittle wing]

so, the tutorial I watched was different from RVC v2 HF?

[fluid latch]

probably

idk which one ur talking abt

[brittle wing]

this but it indonesian: https://www.youtube.com/watch?v=_m5DOgnMI7Q&t=1038s

[fluid latch]

ohhh it is an rvc tutorial

welllll a npy file isnt necessary for an rvc model

[brittle wing]

so, is there any RVC that can work without that file and that is RVC V2?

[fluid latch]

yes all variations with rvc can and will work without an npy file

you only need the pth and index file

[brittle wing]

i see, I hope the RVC I'm using now is one of them

[glad zealot]

are you using the colab pro or free?
are you using the rvcv2 disconnected?

[oblique venture]

Ah nvm i found out how to fix i, thx btw

[rare goblet]

@oblique venture has reached level 1.

[glad zealot]

ah das gud

[oblique venture]

How much epoch should i train for a 5 mins dataset

[potent root]

i recommend about 300-750 epochs

just rely on the graph

lowest point is best point

[oblique venture]

How to check the graph

[patent stirrup]

Any way to use RVC (locally) on mobile

[high night]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English Documentation

https://us.aihubfrance.fr/

Credits: AI HUB France

🇺🇸 English FAQ

https://us.aihubfrance.fr/faq

Credits: AI HUB France

Guides

🍏 Applio-RVC-Fork: Guide

Link: Google Docs

Credits: IA Hispano

How to install RVC on windows

Link: YouTube

Credits: iroaK

RVC v2 Tutorial for Paperspace

Link: Google Docs

Credits: LollenApe

🌐 Translations

• 🇫🇷 French Documentation
• 🇮🇹 Italian Guides or -it_g
• 🇷🇺 Russian FAQ for RVC/Voice changer
• 🇧🇷 RVC Guides in PT-BR (-rvc_br)

[proper shale]

No

[potent root]

tensorboard

[dusty rampart]

i mean if you manage to remote control a computer yes, but with phone processing alone is impossible at the moment

you can also use huggingface spaces

[oblique venture]

I mean, technically you can go to colab and run it on your phone browser

But it not locally so oh well

[patent stirrup]

I taught colab.research.google.com was banned?

[oblique venture]

wait is it? im using it to train my model rn

the v2 one

[violet heron]

Not RVC Disconnected for training

[violet heron]

Do you have stupid

You have the use a online method

You can use colab for training (disconnected is not banned)

For inference you can use wordband or ai-cover-gen

No phone is powerful enough to train

Phones might be powerful enough to inference but we haven’t seen that yet

[turbid vessel]

it seems to be looking for the python shit through "Python\Python39\Scripts" instead of the actual folder its in "Applio-RVC-Fork\runtime\Scripts"

[violet heron]

Hey groovy you need to install torch

[turbid vessel]

what the heck my message got deleted lol

[rare goblet]

@turbid vessel has reached level 1.

[violet heron]

If you edit a message older then 2 minutes it gets deleted

[turbid vessel]

oh ok

[violet heron]

Because someone edited messages to copyrighted stuff and got the server banned

Are you using this for NPCS are becoming smart?

[turbid vessel]

i did that tho and all the other modules manually and it comes back with a new error message

Traceback (most recent call last): File <stdin>, line 1, in <module> File C:\Users\JD\AppData\Local\Programs\Python\Python312\Lib\site-packages\torch\__init__.py, line 443, in <module> raise ImportError(textwrap.dedent( ImportError: Failed to load PyTorch C extensions: It appears that PyTorch has loaded the torch/_C folder of the PyTorch repository rather than the C extensions which are expected in the torch._C namespace. This can occur when using the install workflow. e.g. $ python setup.py install && python -c import torch This error can generally be solved using the develop workflow $ python setup.py develop && python -c import torch # This should succeed or by running Python from a different directory. >>>

[turbid vessel]

could be lol

[turbid vessel]

if anyone else wants to help i said i finished installing applio and got this error message

Traceback (most recent call last):
  File "C:\AI-Training\Applio-RVC-Fork\Applio-RVC-Fork\infer-web.py", line 11, in <module>
    from assets.configs.config import Config
  File "C:\AI-Training\Applio-RVC-Fork\Applio-RVC-Fork\assets\configs\config.py", line 8, in <module>
    import torch
ModuleNotFoundError: No module named 'torch'

[oblique venture]

does the .pth include all of the exist .pth file, or is it only the file with the model name

[violet heron]

that pth file should include the whole pth

[violet heron]

Looks like it couldn’t find the module

[twilit gulch]

so how do I train locally again?

[violet heron]

applio, or the mangio crepe webui

[twilit gulch]

any links?

[violet heron]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English Documentation

https://us.aihubfrance.fr/

Credits: AI HUB France

🇺🇸 English FAQ

https://us.aihubfrance.fr/faq

Credits: AI HUB France

Guides

🍏 Applio-RVC-Fork: Guide

Link: Google Docs

Credits: IA Hispano

How to install RVC on windows

Link: YouTube

Credits: iroaK

RVC v2 Tutorial for Paperspace

Link: Google Docs

Credits: LollenApe

🌐 Translations

• 🇫🇷 French Documentation
• 🇮🇹 Italian Guides or -it_g
• 🇷🇺 Russian FAQ for RVC/Voice changer
• 🇧🇷 RVC Guides in PT-BR (-rvc_br)

[twilit gulch]

thank you

[violet heron]

Dont know a solution , someone else might be able to help

[cobalt wagon]

Im stuck uploading a zip file it won't work. Anyone knows why this is happening?

[patent stirrup]

I inferences gradio often on my phone

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English Documentation

https://us.aihubfrance.fr/

Credits: AI HUB France

🇺🇸 English FAQ

https://us.aihubfrance.fr/faq

Credits: AI HUB France

Guides

🍏 Applio-RVC-Fork: Guide

Link: Google Docs

Credits: IA Hispano

How to install RVC on windows

Link: YouTube

Credits: iroaK

RVC v2 Tutorial for Paperspace

Link: Google Docs

Credits: LollenApe

🌐 Translations

• 🇫🇷 French Documentation
• 🇮🇹 Italian Guides or -it_g
• 🇷🇺 Russian FAQ for RVC/Voice changer
• 🇧🇷 RVC Guides in PT-BR (-rvc_br)

[twilit gulch]

what do I do here

uh
import torch
ModuleNotFoundError: No module named 'torch'

#Server Guide

oops wrong one

#1159513888199540817

[rare goblet]

@hushed bison has reached level 4.

[bold ermine]

Anyone knows how to solve ffmpeg error plz🙏🙏

[twilit gulch]

is this supposed

it has been 10 mins

is it ok?

[wispy burrow]

Never seen that...

[twilit gulch]

is the local applio

I don't like applio anymore

[wispy burrow]

Launch go-web.bat

[twilit gulch]

wrong one brotha

[wispy burrow]

He has mangio.

[twilit gulch]

who is he

oh wait

OH

I thought u replied to me

my bad

[wispy burrow]

Do you still have that error?

[twilit gulch]

im downloading runtime for nvidia

bro

its not even over 10 gigs

[wispy burrow]

Depends on what the error actually is, or what you were trying to do.

[twilit gulch]

💀

[wispy burrow]

For NVIDIA missing 'Torch' module, try to pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117.
That should work.

[twilit gulch]

wadafuk

[wispy burrow]

Is that local or aColab?

[twilit gulch]

ah I did this in python not in cmd bar mb

[wispy burrow]

Surprised that works.
Google usually disconnect people when they run web-UI's.

Never used Colab myself, so I don't know.
Sorry.

Was the link for the model correct?

I think it needs a link that ends with .zip

[twilit gulch]

what is tdqm

[wispy burrow]

Another module not found?

[twilit gulch]

I assume I just do pip install tqdm

[wispy burrow]

correct.

[twilit gulch]

so what do I do to use applio

press 4 or press 1?

[wispy burrow]

1 I'd say.

[twilit gulch]

well 1 gives me an error

[rare goblet]

@twilit gulch has reached level 10.

[wispy burrow]

Maybe try AICovergen.

[wispy burrow]

Module not found errors? Or something else?

[twilit gulch]

I need to download nvidia runtime or sum

???

[wispy burrow]

#1159949278270193734

[rare goblet]

@hushed bison has reached level 5.

[wispy burrow]

Do you have an NVIDIA GPU? Do you CUDA installed?

[twilit gulch]

CUDA yes

nvidia GPU I have an RTX 3060

which is an NVIDIA gpu...

[wispy burrow]

Then what could it be?
Where did you install applio?

[opal relic]

quelqu un peu m aider?

[oblique venture]

If i use RVC-GUI to make an AI song cover, using only cpu (im broke), how long will it usually take?

[wispy burrow]

Depends on the length of the audio, Pitch Extraction method & Feature Retrieval Rate.

[twilit gulch]

desktop

so ima just leave this msg here asking for, anyone got a RELIABLE way to train locally?

[wispy burrow]

Maybe try mangio?

[rigid osprey]

Probably depends on your CPU i'm guessing too. For me a 3 minute song on default settings took around like 5 to 8 minutes

[proud elbow]

make sure you've installed python 3.9 and add its environment path

[twilit gulch]

It’s not that

I already did

[glad zealot]

What does the save model do in applio specifically?

Save model button

[proud elbow]

remove any conflicting python versions and packages?

[twilit gulch]

I don’t know there

[glad zealot]

You can also the one in the pins in this channel but make sure to read the warning

[brittle wing]

i've pressed train index on the colab but it wont train the index

like the index wont appea in my drive

pls help

[stoic cloud]

where did you set the path to save it?

if there is one

It should be somewhere in logs

[stiff goblet]

What is the log interval in Tensorboard? Can someone explain it ? I manually set it to 100. Is it a good value ?

[stoic cloud]

It means how fast it will make a log to update likely

[stiff goblet]

Should i leave it as default or set a manual value ?

[stoic cloud]

Leave it at default, it'll be fine

[stiff goblet]

Oh ok. I actually set it to 100 and training right now. I hope the lowest point will be visible with that value.

[stoic cloud]

Goodluck with training!

[stiff goblet]

Thanks!

[brittle wing]

how do i check the path it's saved to? im on the colab

[stoic cloud]

It should be in rvc/logs/name/added_

[brittle wing]

yeahhh its not working ion know

i think i might try again tmr on local idk

thanksanyway bruv

its alr 3am so

[stoic cloud]

That's a good idea

Cya!

[violet heron]

Yeah but that’s not local

[sand iris]

Should I try local inference on android

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

[candid dust]

there isnt a bugs section etc just want to post that rcv disconnected notebook still is broken as far as the pitch guidance goes it still trains with guidance even if u turn it off

[swift merlin]

hey guys, quick question, from outside the gui, what's the folder to drop my .pth and .index files in applio rvc? It seems they're on the log folder. Is this permanent?

[brittle wing]

How can I use a premade voice models on RVC2 local?

I can’t find anything about rather than creating datasets and voice training.

[rare goblet]

@brittle wing has reached level 2.

[rigid osprey]

What's RVC2 local? Like where did you download it?

[zinc anchor]

I think they just meant RVC V2

[wispy burrow]

ModelName.pth should go in logs/weights.
added_IVF*.index should go in logs.

[wispy burrow]

Download a model from #1159289700490694666 or weights.gg.
Place the ModelName.pth the folder 'weights'.
Place added_IVF*.index in the folder 'logs'.
Then you can use the model for Inference.

[spiral osprey]

my disconnected rvc just had this problem, can anyone help me? I had put the zip in the wrong place, then I put it in the right place and then it didn't recognize it, then I reloaded the website and had this problem, can anyone help me? I'm trying to make a model there

[astral perch]

needs more training right?

[brittle wing]

can anyone help me with this? i have python installed with the Add to path checked

[brittle wing]

What about voice models that I have trained myself?

[rigid osprey]

Try running "python --version" what does it say

[wispy burrow]

Those should be in those folders as well.

[rigid osprey]

Can you rename index files? They always have long filenames, those don't matter right?

[brittle wing]

Very much, appreciated it.

Thank you

[brittle wing]

3.9

[rare goblet]

@brittle wing has reached level 2.

[nocturne mural]

what it does is to save your model in a .zip file

[modest spruce]

Is there any way to get local to train as fast a colab?

[wispy burrow]

You can.

[daring verge]

get better gpu

free up vram

avoid running background

[stoic cloud]

Get a good GPU

[brittle wing]

Get the Nvidia h100 gpu

It will be fast as colab

No cap ong

[daring verge]

[oblique venture]

Anyway to look into the graph without downloading RVC onto my computer? Like is there a colab one or anything that doesnt require downloading RVC

[daring verge]

you mean tensorboard?

[oblique venture]

Like, i can use tensorboard, but to look into the graph the guide said it require download rvc onto my computer, for certain reason i cant download it

[daring verge]

let me see that guide

[proper shale]

I think you just need the logs folder

And TensorBoard

[oblique venture]

https://youtu.be/P0M7PAsG1fk?si=jcldrKCw3QyRXzY-

[oblique venture]

How to put the log folder into tensorboard 🤔

[proper shale]

Nah you just need to open the events file

iirc

[atomic dock]

is RVC-HF, by r3gf, listed in the info tab a good place to make the ai covers?

[rigid osprey]

Did you fix it?

[shut marsh]

"Cannot connect to GPU backend" Didn't get to finish training

[burnt zealot]

can i train one voice using both gpus at the sametime using the option i circled

[tribal badge]

-colab

[azure marsh]

Google Colabs

RVC

• AI-Cover-Gen-No-UI (inference only)
• RVC v2 Disconnected (training only)

🌐 Translated colabs

• PT-BR: -colabs_br

Audio Separation/Isolation

• MDX Net Colab
• UVR arch colab 5.0.2/c1.04
• KaraFan colab

[tribal badge]

IS VERY ANY WORKING RVC

[brittle wing]

Idk how tf it works lmao can you record a video of it using a virtual windows or a new window that have no python in it? Shits confusing

[rigid osprey]

Idk how to fix that error either :(

[brittle wing]

Damn

[brittle wing]

-rvc

[azure marsh]

RVC Guides (How to Make AI Cover)

Documentation

🇺🇸 English Documentation

https://us.aihubfrance.fr/

Credits: AI HUB France

🇺🇸 English FAQ

https://us.aihubfrance.fr/faq

Credits: AI HUB France

Guides

🍏 Applio-RVC-Fork: Guide

Link: Google Docs

Credits: IA Hispano

How to install RVC on windows

Link: YouTube

Credits: iroaK

RVC v2 Tutorial for Paperspace

Link: Google Docs

Credits: LollenApe

🌐 Translations

• 🇫🇷 French Documentation
• 🇮🇹 Italian Guides or -it_g
• 🇷🇺 Russian FAQ for RVC/Voice changer
• 🇧🇷 RVC Guides in PT-BR (-rvc_br)

[brittle wing]

this TTS can use custom voice models?

[rare goblet]

@brittle wing has reached level 3.

[brittle wing]

any idea why applio still cant find my dataset?

[oblique venture]

Guys are there anything that like RVC-GUI but on colab, my pc only has cpu and it taking ages to convert a song 😅

[brittle wing]

outdated but it should work
https://colab.research.google.com/drive/1TU-kkQWVf-PLO_hSa2QCMZS1XF5xVHqs?usp=sharing

oh wait no

it doesnt

cuz they banned gradio

sorry idts

[oblique venture]

Ah ok

Bummer

[brittle wing]

yeah sorry

but i'm not working on anything rn cuz i cant figure out how to train, you could outsource what you want to do to me?

[strong belfry]

rvc disconnected colab

[oblique venture]

Thank for the offer but i prefer keeping my stuff private

[rare goblet]

@oblique venture has reached level 3.

[oblique venture]

Lol i mean create AI song cover, i already got my own model 💀

[strong belfry]

you use the same things to convert and train voices?

[brittle wing]

any idea why applio can't find my dataset?

[strong belfry]

it should be in an another folder inside the dataset folder

then click refresh

should then popup

[brittle wing]

AHHH perfect i thjought it was suppsoed to just be the wav, thanks heaps!

[rare goblet]

@brittle wing has reached level 4.

[brittle wing]

how do i find my gpu indexes?

[strong belfry]

you shouldnt need to

only if you have multiple gpu

[brittle wing]

gives me this error, though

[oblique venture]

I thought the rvc disconnected only train voice model, not create AI song cover

[brittle wing]

but the preprocessing step worked

[violet heron]

Local

[strong belfry]

are you sure you processed the files? there should be a lot of wavs in 1_16k_wavs folder

[oblique venture]

Guess im gonna wait 10 years+, see u guys later

[strong belfry]

thats also a vague error. what does the terminal window say?

[violet heron]

Wordband

If you really don’t want to use local for some odd reason

[strong belfry]

can you not make inference on rvc disconnected?

[rare goblet]

@strong belfry has reached level 6.

[violet heron]

What’s wrong with local anyways?

[brittle wing]

i don't have that folder i think

[strong belfry]

then the step 1 isnt done

[violet heron]

CPU runs fine on infrencing

[strong belfry]

depends on the cpu

[violet heron]

True

[brittle wing]

says it's done

[strong belfry]

might not even have enough ram if you dont have a gpu

[violet heron]

I have a office computer

[strong belfry]

its done beacuse there is an error

[violet heron]

Takes no more then 10 minuted

[brittle wing]

oh

[violet heron]

ITS A 2ND GEN i5 FOR CRYING OUT LOUD

WITH ONLY 8 GIGS

[strong belfry]

ffmpeg errored out. your input dataset isnt processed

double check your dataset is just a folder of wavs inside the dataset folder

[oblique venture]

I mean i dont really have any problem with local, i actually prefer it, the only problem is the time

[brittle wing]

should be

[violet heron]

You can try wordband, it’s a bot in the server which can create covers

[oblique venture]

Mine is i3-6006U 4gb 💀💀

[strong belfry]

how is that wav encoded?

[violet heron]

You can register a model from HuggingFace from /register

[brittle wing]

it was a podcast in audacity and i just chopped up the bits where the guy i wanted was speaking

[strong belfry]

how long is it?

[brittle wing]

5 minutes and 1 second

[strong belfry]

thats odd

[brittle wing]

do i need to manually install ffmpeg or something

[violet heron]

https://docs.google.com/document/d/1-1t1iQDST5dVExejFnfyZaF_qUwcgqmYqzGOjryyXrQ if you need a guide for it

[strong belfry]

i dont think so

[brittle wing]

hmm

[oblique venture]

Thank mate

[strong belfry]

process the file again, then show full error in terminal?

[brittle wing]

Hi guys

how can I know if it's overtrained or nah on tensorboard

who knows how to use it

[strong belfry]

-overtrained

ah

[violet heron]

Since your using a model you created you would register it

[strong belfry]

i have no idea how the view the full error. its not printing all of the errors that ffmpeg had

try exporting the wav again

make sure its mono, a reasonable samplerate (40-48k)

[brittle wing]

ohhh it weasnt mono hold on

[strong belfry]

its shouldnt error out if its isnt mono

[brittle wing]

this correct?

[strong belfry]

that looks fine

[brittle wing]

Okay

let me try

[strong belfry]

try exporting it as a flac or mp3 if it still doesnt work

16bit flac or 320kb mp3

[brittle wing]

y'all just miss my question 😕

[strong belfry]

That should feed into the voice model

[brittle wing]

still doesnt work :(

is it possible it isnt a problem with the dataset?

or does the error suggest it is

[brittle wing]

how, I just want text-to-speech on purpose, any method?

OHMYFOD I FEEL SO STUPID

because i enterrd the custom

directory

before making the folder

it tried to process the folde rnot the dataset

sorry for wasting your time dolyfin

thanks so much

[strong belfry]

ah i see

[strong belfry]

just read the instructions there? leave blank for free google tts or pay elevenlabs for tts

[brittle wing]

i really don't know what's that mean because i'm new on this

[strong belfry]

type in some text and press speak?

[brittle wing]

no, I know that but what I mean is whether it can use custom voice models for it? Not the default one

[strong belfry]

You are supposed to use a default tts, then use the voice model to convert the default voice into the voice model's voice

the rvc model is not a text to speech model

its a speech to speech model

[brittle wing]

that mean it only for clone voice?

[strong belfry]

its needs a base voice to work with

which is why you would use google tts to generate a voice which you can then convert

[brittle wing]

i see, I hope I find the right tutorial for it, the Ai Cover tutorial is not what I want

[daring verge]

you can get the API for elevenlabs for free though

[glad zealot]

In which folder does it save it to?

[oblique venture]

@violet heron holy the wordband bot worked, thank you!

[coarse crater]

ey guys does anyone have the link to rvc v2 google colab?

[oblique venture]

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=Q7JhsK5oYByc

[merry wharf]

On RVCconnectred colab, is there any way to SAVE all the epochs from the wehights folder to the google drive?

[fathom olive]

you can make a new cell with this command - "cp -r /content/Mangio-RVC-Fork/weights /content/drive/MyDrive/"

Another question, why RVCDisconnected does not adapt the sample length when cutting audio for GPU? 3.0 seconds per sample instead of 3.7 as it was before. Is this an optimization? What is the rationale behind it?

[lean owl]

will there ever be an rvc v3

[fathom olive]

In the process of testing at the moment

[jaunty silo]

i thought it got canceled

[loud summit]

One message removed from a suspended account.

[lean owl]

how much ram do u have

is it like an 8gb ram laptop

or 4gb

or whatever

idk

[loud summit]

One message removed from a suspended account.

One message removed from a suspended account.

[lean owl]

i think thats pretty good

[sand iris]

Can't train on that though

But (ab)using google's services is always an option

-colab

[azure marsh]

Google Colabs

RVC

• AI-Cover-Gen-No-UI (inference only)
• RVC v2 Disconnected (training only)

🌐 Translated colabs

• PT-BR: -colabs_br

Audio Separation/Isolation

• MDX Net Colab
• UVR arch colab 5.0.2/c1.04
• KaraFan colab

[daring verge]

just use rejekts colab

[loud summit]

One message removed from a suspended account.

[rare goblet]

@loud summit has reached level 2.

[loud summit]

One message removed from a suspended account.

[proper shale]

Dataset is not on the rvcDisconnected folder on Google Drive or the zip isnt formatted correctly

[loud summit]

One message removed from a suspended account.

[loud summit]

One message removed from a suspended account.

[proper shale]

The dataset files must be in a folder before you zip it

[loud summit]

One message removed from a suspended account.

[proper shale]

So your zip will extract as a folder with audios

[proper shale]

Yes

[loud summit]

One message removed from a suspended account.

[proper shale]

Nope, not rn

[loud summit]

One message removed from a suspended account.

[proper shale]

But basically, zip the folder the audios are in.

Got it?

[loud summit]

One message removed from a suspended account.

[proper shale]

Like this:

You have a dataset for "mario"

It should be a folder named "mario" formatted like this:

• mario (folder)
• audio1
• audio2

[loud summit]

One message removed from a suspended account.

[loud summit]

One message removed from a suspended account.

One message removed from a suspended account.

[grim bay]

Is there a guide for this?
https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb

[grim bay]

I need a guide for this

[oblique venture]

If i use two different voice as dataset, then the model gonna learn both of the voice right, like some type of mixing? Or does it work in another way

[proper shale]

it'll make your model super weird probably

[grim bay]

How can I upload my files to this Goggle Colab link?
https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb

[rare goblet]

@grim bay has reached level 2.

[grim bay]

Help please

[proper shale]

There's a file explorer on the left side

Use that to upload files

Also there's a WebUI version that works by Hina

Check pins

[grand willow]

when to stop training?

[loud summit]

One message removed from a suspended account.

One message removed from a suspended account.

[proper shale]

Have you done feature extraction

[loud summit]

One message removed from a suspended account.

[rare goblet]

@loud summit has reached level 3.

[proper shale]

...the second most important step in training

[proper shale]

decrease smoothing to 0.89 or something

[loud summit]

One message removed from a suspended account.

[proper shale]

...have you followed all the steps in order?

[loud summit]

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

One message removed from a suspended account.

[proper shale]

oh...

hmmm

[loud summit]

One message removed from a suspended account.

[proper shale]

then you messed up your zip again i think

because

the folder's name inside the zip must be the same as the character name you set in the training variables

[proper shale]

@rigid mango is this how it should be? now I'm curious

[loud summit]

One message removed from a suspended account.

One message removed from a suspended account.

[grim bay]

https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb
Dang it, this doesn't work anymore

[proper shale]

how's the d/total

[proper shale]

Use the one in pins

[stoic cloud]

2K is reffering to the steps right?

[proper shale]

yes.

[stoic cloud]

@grand willow how many wavs did you have in your dataset and what's your batch size

[rare goblet]

@grand willow has reached level 1.

[proper shale]

i think it might be overtraining

[stoic cloud]

How many wavs do you have and how big is your dataset

[glad zealot]

dang people can still use tensorboard

wish i can (on my custom)

[stoic cloud]

I don't access tensorboard

I could though

[glad zealot]

teach me how to implement it OwO

[grand willow]

one big wav 23 minutes made from 7 songs also delete any Spaces using audacity

[stoic cloud]

It should just be in the folder somewhere

[glad zealot]

heres my code to run it stolen from rvcv2disconnected

[stoic cloud]

Songs?? Is there background music in there?

[grand willow]

nooooooooo

[glad zealot]

tensorboard?

[stoic cloud]

Yeah

[grand willow]

it is pure singer tracks i know how to isolate but it is a tracks from my studio

[glad zealot]

im using the applio mangio fork

the one that disconencted was using before

[stoic cloud]

It would have been beneficial to cut the audio files up into smaller chunks of 10 seconds each

[grand willow]

i made many special models before but i always can,t decide when to stop training

[stoic cloud]

How does it perform

The model that you are training

In the tensorboard you should be able to see

[grand willow]

i sent the photos but i still never find the point to stop

[rare goblet]

@grand willow has reached level 2.

[glad zealot]

@stoic cloudye idk how set up tensorboard with this

[stoic cloud]

I use local

[glad zealot]

reeeeeeee

[stoic cloud]

Yeahhh

[glad zealot]

it does fully work tho

just no tensorboard

sadge

[strong belfry]

I think you could just copy out logs folder to your local pc. install tensorboard, then point it at the logs folder

not sure how to copy things out of colab

[glad zealot]

download so no

[strong belfry]

can you not download files out of colab?

[glad zealot]

i mean i can download them but it takes a while to download

like 15-30 sec every single file

also need to auto move them to the correct folders in local so ye... dont think so

actually i saw the tensorboard file lemme try accessing it in colab

[grand willow]

use g/total or d/total to know the point to stop training in tensoboard ????

[proud elbow]

-overtraining

[azure marsh]

How do I know if my model is overtaining?

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

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

Credits: LUSBERT

Automated Overtraining Detection (AOD)

Will be available soon in #1159513888199540817

Credits: grvyscale

[grand willow]

g/total

[proper shale]

both

[grand willow]

d/total

[proper shale]

I'd say try out the model at 5k steps

And at around 8-9k

[grand willow]

thanks

[wooden lodge]

Someone help me make ai covers

[tame mica]

uh

[astral perch]

Some point on the graph look lower but other points on the graph have lower value but look higher

[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!

This isn't an RVC2 model! Please make sure to use an RVC2 model from AI Hub. (Make sure the model is tagged with RVC v2)

Successfully registered model control (scp containment breach mtf control)!

Use the model now like this:
/cover model:control youtube:https://www.youtube.com/watch?v=s9FCCiEs84k

[green birch]

do i have to have wav only? and how long or big should files be i want to train offline cuz it just seems better for my needs but im stuck on the train part i have some ref voice stuff sitting around but what to do next is where i get confused (pls not this is NOT for singing its real time voice changer ) as such i cant seem to figure it out ping relpy or whatever

if this is online only its sorta a issue

[plucky locust]

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

[modern snow]

Les go front of the cue

[rare goblet]

@modern snow has reached level 4.

[torpid depot]

Guys how do I fix the unhandled rejection: error stack problem

[rare goblet]

@torpid depot has reached level 1.

[spare phoenix]

Hello, I wanted to know how to make an ai cover? I haven't made one since the old collab shut down

[rare goblet]

@spare phoenix has reached level 1.

[sand iris]

It should support just about any audio format

[brittle wing]

what tf is mean? Runtime Disconnected and it always happened
any solution?

[sand iris]

Wait what do you want to achieve? Convert a voice or make a model ?

And which software are you using

[green birch]

make one from scratch offline thats the thing im brand new i have 0 idea i took video and pulled voices out but got stuck there

[glad zealot]

that seems to be blocked by google so use the rvc discoonected one

its in #1159949278270193734

[brittle wing]

different from what i saw on YouTube but i will try

[rare goblet]

@brittle wing has reached level 4.

[brittle wing]

btw this rvc discoonected has a tutorial?

[glad zealot]

not that i think of

[proper shale]

yes it does

[half cove]

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

[proper shale]

#1159513888199540817

Also I'm making one

[glad zealot]

ahh there we go

[brittle wing]

thanks

[brittle wing]

what its mean by "Upload the dataset zip file into the folder"?

[proper shale]

upload your zipped dataset to the rvcDisconnected foldef

[brittle wing]

which file type?

[glad zealot]

zip

just make a folder called rvcDisconnected

then inside just put the zip that you want to train

[brittle wing]

ahh, i see

[half cove]

In the disconnected folder in your Google drive

Oh shi i was late

[glad zealot]

your zip file must have a folder inside of it

inside that folder is your dataset(s)

[torpid depot]

guys everytime I select an input in the rvc client it gives me a unhandled rejection error stack crash what do I do to fix this exactly

[glad zealot]

what browser are you using?

also #🔍│help-w-okada

[torpid depot]

oh alr

[green birch]

the end result is i want something for real time voice , i just donno how to do it and would pref offline if possible so i can do my gpu

moving to okada i think im in the wrong room as well

[torpid depot]

wdym by this btw because like isnt the client seperate

[glad zealot]

ahhh its electron

i see

just try reloading it if you cant

ctrl+f5 probably

[torpid depot]

yeah I tried reloading, initializing, restarting my pc

same problem

[brittle wing]

where should I find this?

[glad zealot]

wut

[astral perch]

Some points on the graph look like the lowest point but have a higher value than other higher points

[oblique venture]

If i change the model at hugging face, does the model on wordband change too, or does it stay as the old model?

[brittle wing]

i mean, what does it mean "Delete the path in "Load Dataset" cell?

[oblique venture]

Replace the zipname.zip with your dataset file name

[rare goblet]

@oblique venture has reached level 4.

[oblique venture]

For example, zipname.zip -> whateveryourdatasetnameis.zip

[brittle wing]

i found it, thanks

[oblique venture]

Np

[grim bay]

Are there any Goggle Colab links except this?
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb

I want to use the one that the AI doesn't separate instrumental and vocal

[glad zealot]

wdym by doesnt separate?

[twilit gulch]

just wanted to ask, what is the mangio trainer link?

[grim bay]

The old one doesn't have the UVR in the AI

I want the old one

Sometimes that Google Colab AI link just separate the vocal even thought I did that on the UVR

Making the result is kinda bad

[glad zealot]

so are you trying to get just the vocal and just the music?

[grim bay]

Not really
I want to make the AI covers

But like I want to use the vocals only because I used to do that a lot

[glad zealot]

just get the vocal part of the cover?

ye?

[brittle wing]

i got this error, what i should to do?

[grim bay]

I have the vocals but it still trying to separate again when I use the AI

I want an AI that just process the files and not separate it

[glad zealot]

the vocal only is in the files

you can also just download the full cover (voice and music)

why are you making this more painful for yourslef

[grim bay]

Because I tested the AICoverGen and the Magio
And the result is the one that I separate by using UVR and the Magio to process the singing gives me the better result

[rare goblet]

@grim bay has reached level 3.

[glad zealot]

just download the full cover version then

[grim bay]

The result is not as good
Especially when I use the high pitches song

[glad zealot]

ye not sure about that then

that one is from the original , if the original updates, that also updates

[glad zealot]

is this the one that you want?

[grim bay]

Yes, that one
I remember there is an easy UI version

[rigid osprey]

Has anyone tried elevenlabs vs coqui.ai? I'm not sure which is better/faster

[oblique venture]

If i change the model at hugging face, does the model on wordband change too, or does it stay as the old model?

[twilit gulch]

mangio link?

[glad zealot]

ye sadly i cant share this colab one yet sorry

[grim bay]

Oh, why is that if you don't mind

[twilit gulch]

gatekeeping:

/wip:

[brittle wing]

can u explain this? i need help

[twilit gulch]

show full error

[glad zealot]

currently on testing and google doesnt really like people using webui in colab so if lots of people use it , ti will get patched

[grim bay]

Oh

[brittle wing]

python train_nsf_sim_cache_sid_load_pretrain.py -e "Homura" -sr 40k -f0 1 -bs 8 -g 0 -te 200 -se 25 -pg pretrained_v2/f0G40k.pth -pd pretrained_v2/f0D40k.pth -l 1 -c 0 -sw 1 -v v2 -li 1

FileNotFoundError Traceback (most recent call last)
<ipython-input-53-32820e447018> in <cell line: 82>()
80 f0nsf_dir = "%s/2b-f0nsf" % (exp_dir)
81 names = (
---> 82 set([name.split(".")[0] for name in os.listdir(gt_wavs_dir)])
83 & set([name.split(".")[0] for name in os.listdir(feature_dir)])
84 & set([name.split(".")[0] for name in os.listdir(f0_dir)])

FileNotFoundError: [Errno 2] No such file or directory: '/content/Mangio-RVC-Fork/logs/Homura/0_gt_wavs'

[twilit gulch]

hmmm...I don't know!

[mint stag]

is it possible to pause the training and continue where it left off later?

[grim bay]

https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing
This one is the one I used to use
But it's not working anymore

[glad zealot]

yes

-colab

[azure marsh]

Google Colabs

RVC

• AI-Cover-Gen-No-UI (inference only)
• RVC v2 Disconnected (training only)

🌐 Translated colabs

• PT-BR: -colabs_br

Audio Separation/Isolation

• MDX Net Colab
• UVR arch colab 5.0.2/c1.04
• KaraFan colab

[mint stag]

so I close the cmd that's training and then when I want to restart training I just open the gui and put the same info and press train only right?

[glad zealot]

local?

[mint stag]

Yes

[glad zealot]

if local you can just stop training and close it

then just resume later

[mint stag]

yeah and to resume I need to reopen the web gui and press train right?

and it will continue where it left off?

anyone know how to fix tensorboard being bugged?

[proud elbow]

is it mangio? and did it show nan values during training?

[mint stag]

Yes, mangio.

yes nan

[proud elbow]

what's your GPU? I had that same issue when I had an old 6 GB GPU, but then I upgraded to a 12 GB GPU and it worked properly. so I can assume it needs at least 8 GB VRAM.

[mint stag]

gtx 1650, but I saw other people already being able to train on it

and I used a batch size of 2, my vram is 4

[proud elbow]

but my old GPU worked on the original RVC and applio for training though

[mint stag]

so it's most likely not a gpu issue then

[brittle wing]

so what app to make clone voice using custom voice model?

[rigid osprey]

Probably this one: https://github.com/Tiger14n/RVC-GUI

Maybe this one if you want real time: https://youtu.be/xwdbwbtO-FQ

[proud elbow]

or it could probably be my driver which was either 536.67 or 537.13 alongside my old GPU (GTX 1660 super). after I have my new GPU (RTX 4070) and I have 537.42 and also rolled back to 531.79, it works properly.

[brittle wing]

i'm using windows 7

[modern snow]

A 4070? Nice

[modern snow]

Ok? Why

[brittle wing]

my life is hard

[mint stag]

Oh I see. Alright I will try updating/rolling back my driver, thanks

[modern snow]

Oh. Yeah I feel you there my best device for making models is my phone

[rigid osprey]

@brittle wing Not sure what will/won't work with windows 7. Try looking in guides it might have more information

[brittle wing]

i have try that long ago #🔍│help-w-okada message

[brittle wing]

I just want to find a way to change someone's voice to Ai's voice with using custom voice models, that's all

[glad zealot]

does it disconnect after 5 mins or something?

[rare goblet]

@glad zealot has reached level 23.

[rigid osprey]

Most of this stuff is pretty new so I doubt much will have support for windows 7. Maybe there is an online service you could find, not sure

[brittle wing]

if so, i gonna proud of myself

[grim bay]

Yeah

[twilit gulch]

mangio download?

[wispy burrow]

Looking for Mangio?

[twilit gulch]

nvm already downloaded

how do I check for overtraining tho

with a local tensorboard or sum

I got the D** and G** files

brotha please im already reaching 500 😭

[wispy burrow]

Installing TensorBoard locally.
https://docs.google.com/document/d/1ihvqWnSyabvZ9jX6BsOxEY1FpplwjY3Jse3she_VO6I/edit

[rare goblet]

@wispy burrow has reached level 14.

[glad zealot]

u want it to work?

:D

[twilit gulch]

to stop training

do I jsut close the command prompt

[wispy burrow]

Doesn't the training button say "Stop Training"?

[twilit gulch]

a

and if I press train again will it resume?

[wispy burrow]

It will if you keep the Experiment settings the same and if you keep the files in logs.

[twilit gulch]

so basically to resume progress

screenshot all of my things

reinput them when I open my thing so it's the same values

and just click training again (if I didn't change the raw files ofc)

and change these 2 too?

[rare goblet]

@brittle wing has reached level 5.

[grim bay]

Yeah, but I don't want to upgrade to Pro

[wispy burrow]

Everything the same.
You can change the Total Epochs, save frequency and such though.
No need to process the Data or extract Feaures again.

[proud elbow]

Clicking "Stop Training" will continue finishing the current epoch and then save the model checkpoint at the epoch, and it's safer way than force closing the application.
to resume training, you can do it when starting the application again, as long you keep the current G and D files and use the same batch size and total epoch more than the current.

[twilit gulch]

thank you, both of you

I now have the knowledge to train all of my models thanks to you

[pine grail]

help what is download checks.json

[glad zealot]

currently trying to make it work

on free

[glad zealot]

i make it slightly work? can do 2 song covers every time before it dies

[twilit gulch]

hey uh

in the pretrained base model path (D AND G)

what context do I give the thing

[wispy burrow]

No idea on your issue but I notice there is a space in the path to the RVC folder.
It is recommended not to have any spaces and special characters.
Those may cause issues.

[twilit gulch]

do I do ./?
do I just do \?
do I do my full path from C:?

ffs I can't type \

[wispy burrow]

Aren't those loaded automatically?

[twilit gulch]

nono

the pretrained ones

[pine grail]

OH MY GOD I ALREADY CHANGED MY USERNAME AND THE USERNAME HASNT CONVERTED INTO THE FILE EXPLORER THING

[grim bay]

You mean that link?

[brittle wing]

why when i start RVC-GUI this massage appear, any solution?

[twilit gulch]

uprade pyth

[glad zealot]

before colab terminates it

[twilit gulch]

man this tensor board stuff didn't work locally, does anyone know how to check the files locally

as in, how to check progress on the epoch thing (training)

[brittle wing]

where i should find the latest one?

[twilit gulch]

https://www.python.org/

https://www.python.org/downloads/release/python-3120/

[brittle wing]

it need Python 3.8.0

[twilit gulch]

then download python 3.8.0

[brittle wing]

looks like i can't use that app...

so, any app that can make clone voice using custom voice model than it?

[twilit gulch]

so how do I check my overtraining progress locally??? 💀

brotha please 😭

[proper shale]

You install TensorBoard

if you haven't yet

[twilit gulch]

and what do I do with said tensorboard

[proper shale]

Analyze the graphs

[twilit gulch]

put it in a way where I can understand

[proper shale]

Mainly g/total and d/total

You'll look for the lowest point in g/total basically

[twilit gulch]

do I open a 2nd command prompt to view said information?

how do I view the D_ and G_

oh jesus fucking christ

[proper shale]

Yeah

[twilit gulch]

D** and G**

[proper shale]

Scalars tab, search for total

Decrease smoothing if you're training for long enough

[twilit gulch]

do I need to write a piece of code to open the data w tensor?

[proper shale]

no, just launch tbL.bat

[twilit gulch]

I know how to use the board itself I just don't know how to open it

[twilit gulch]

will it open a thing on google

[proper shale]

The browser? Yes

That's the TensorBoard page

[twilit gulch]

holy fucking

thank you so much

that's so simple wtf?

I give you biggest thanks.

[proper shale]

npnp

pleasure to help! :)

[brittle wing]

please, i have try everything but some that I tried seemed to refuse me to do it, i just really need it 🥲

[twilit gulch]

so now I just keep going w my echops until I see it go up?

Epochs

[proper shale]

Well, analyze it like this - if it starts going continuously up and never goes back down after a while

that's how you know you're overtraining

And then you select the lowest points before this overtraining

[proper shale]

Can you elaborate on that question - I'm not sure I understand

[twilit gulch]

but mine saves every 50 and doesn't save the ones behind it

I think...

I don't know yet ! I'll ask if I need help

[proper shale]

that's tough

should've used 25 or 20 at least

[twilit gulch]

can I pause it, change the value and ocntinue?

[brittle wing]

i need to find a clone voice that can use custom voice models then i have try every app but all of it It can't be opened and it doesn't fit what I want, windows 7 really sick me up! 🥲

[proper shale]

Depends on what GUI you're using

[twilit gulch]

mangio

local

[proper shale]

latest release?

[twilit gulch]

the one in info

I downloaded 2day

[proper shale]

ohh

then the "Stop training" button should be there

will stop on the latest checkpoint

[surreal delta]

@ everyone how to make ai cover

[proper shale]

Give me nuclear launch codes

And Ill gladly walk you through it

[vast kraken]

do you still get the models link from drive

[proper shale]

No, Huggingface is used more now

[vast kraken]

how will i get them to work

[rare goblet]

@vast kraken has reached level 1.

[vast kraken]

like

how do i download them on gradio

[proper shale]

what fork r u using

[vast kraken]

i'm using ai cover gen web UI and i want to download a voice model. how will i do it

like this how will i get link

[proper shale]

in voice models

or weights.gg

[twilit gulch]

do I stop soon

so if I trained to 2k epochs

what do I use for inferring voice

[proper shale]

check scalars tab

[proper shale]

use the latest

[twilit gulch]

wher

do I see that

this shit buggin

well thanks again

what the fuck happened

[proper shale]

oh yeah

it didnt train properly.

[twilit gulch]

what

2 hours down the drain?

welp there goes my free time for this week !

[proven sapphire]

/search travis scott

[ionic adder]

1 result found

travis-scott

• Author: huggingartists
• Downloads: 18
• Likes: 0
• Tags:

["transformers","pytorch","jax","gpt2","text-generation","huggingartists","lyrics","lm-head","causal-lm","en","dataset:huggingartists/travis-scott","endpoints_compatible","text-generation-inference","region:us"]```
- **Pipeline-Tag:** `text-generation`
- [**URL**](https://huggingface.co/huggingartists/travis-scott): https://huggingface.co/huggingartists/travis-scott

[proven sapphire]

this is shit

[violet heron]

np

[rare goblet]

@violet heron has reached level 30.

[violet heron]

Wrong bot

[shy zealot]

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

Travis Scott

Upload the <t:1694426487>
Server uploaded: AI HUB (RIP)
Channel uploaded: AI HUB (RIP) Post
Profile Owner: starboyz_z

Author

StarBoy

Discord User

starboyz_z

Internal ID

8262

Language

EN 🇺🇸

Type

N/A

Size

0 MB

Extractor

N/A

Epoch

0

Steps

0

Powered by ModelAI.net, Made with Love ❤️

[twilit gulch]

So what do I do?

[rare goblet]

@twilit gulch has reached level 11.

[loud summit]

One message removed from a suspended account.

[azure marsh]

Google Colabs

RVC

• AI-Cover-Gen-No-UI (inference only)
• RVC v2 Disconnected (training only)

🌐 Translated colabs

• PT-BR: -colabs_br

Audio Separation/Isolation

• MDX Net Colab
• UVR arch colab 5.0.2/c1.04
• KaraFan colab

[proper shale]

well, try again

you probably screwed up on some part of it

[twilit gulch]

I?

Me?

[proper shale]

happened the first time i trained so

[twilit gulch]

I obeyed every single tutorial, step, everything.

[proper shale]

errors can happen

¯_(ツ)_/¯

[twilit gulch]

Sigh, just send me the mangio rvc download link so I’ll redownload

And do everything again

[proper shale]

no you don't need to redownload everything

[twilit gulch]

But I want to.

[proper shale]

...why?

[twilit gulch]

Why not

[proper shale]

at that point you'd be just wasting more time

[twilit gulch]

Maybe I downloaded the wrong version or some shit

Who cares about time at THIS point

Literally just want a shitty AI that won’t probably be that good

[proper shale]

Apparently your feature extraction went wrong

[twilit gulch]

What

Really?

[proper shale]

At least that's what I could infer from the error you had

So maybe the files in your dataset were put in the wrong way (not extracted correctly or in a filetype ffmpeg doesn't support)

[twilit gulch]

Nah

It’s all WAV and OGG

[proper shale]

How was it placed in the dataset folder

oh that too

[twilit gulch]

Aghhh

I see

so…I need to grab 400 files and change their file type…fuck

[twilit gulch]

C:/blablabla/Downloads/Cell/Voices

Wdym

I can just input the voicelines as 1 line?

[proper shale]

i do this

[twilit gulch]

Really?!

So I can just input it all into audacity, change position 1 after another, export as wav and it works?

So FLAC

[proper shale]

...wav is better for compatibility

but eh you do you ig

[twilit gulch]

Flac less space

._.

still supported by ffmpeg

What if it OGG and wav

What

But no oggs is like…removing 50%

Brother, from 25 mins it goes to 10

thank you for your knowledge

I will be making 2 models, one OGG, one WAV
And I will be comparing them

[vague sage]

whats the best drive michael jackson model?

[proper shale]

mine

(its huggingface)

[twilit gulch]

I used video game lines

[vague sage]

qual é o seu?

[proper shale]

[twilit gulch]

(I have been working on voices nd extraction for 2 years know)

[proper shale]

o da era Invincible, 500 epochs

[vague sage]

tem muito drive? aquele raspado na voz?

[proper shale]

ss

especialmente qdo a nota é alta

[vague sage]

top

[rare goblet]

@vague sage has reached level 2.

[vague sage]

vou testar

[vague sage]

https://drive.google.com/file/d/1w9UR6gTTZG1EAw3gAGUCwJqmmN-feVeB/view esse?

[rancid knoll]

yo does anyone know

where i can download this model?

i couldn't find a discord server for UVR so i came here

[proper shale]

nope, esse é ruim, o meu é de 500 epochs

[vague sage]

n tem o seu

[proper shale]

e tem um dataset bom

[vague sage]

manda aq o link

n tem em modelos

[proper shale]

tem no weights gg

[vague sage]

onde é isso?

[proper shale]

https://weights.gg

[vague sage]

aah top

n sabia q tinha isso

[rugged stirrup]

Does anyone know a python script that splits audio

[proper shale]

why would you want to split audio

[rugged stirrup]

For a dataset why not

[proper shale]

because rvc automatically does that

unless its a really big file (1 hour)

[rugged stirrup]

well for long datasets it doesnt train the index

[proper shale]

well you could just cut off every 30 mins

lol

[astral perch]

Do you have to look at the lowest value on the graph

[proper shale]

Yes

[rancid knoll]

yeah

[astral perch]

Because some lower points have higher values than higher points wich confused me

[proper shale]

But remember to always check every 50 or so epochs

[rancid knoll]

that's the effect of smoothing on the graph

[proper shale]

And check those points too