#WILLING TO PAY FOR A FULL TUTORIAL ON RVC
1 messages · Page 1 of 1 (latest)
do you have reading skill issue on the guide? https://docs.aihub.wtf/
Last update: Mar 10, 2024
The guides have everything you need for free
The guides are not it
And if you are willing to pay at this point you might as well pay for a model
And
no rx in depth stuff
They're actually goated
i don't want to pay for a model i want to become a master myself
so i need to find one
are you willing to ?
Then look at the guides
I;ve been looking at them for more than a year
Not really I used the guides
They didn't help me enough as some admins in here
And they've worked
You know that there is some sauce that the QC's dont give out 100%
every competitive field has it
im trying to find it
the guides cover basic stuff
Start with the basics
i already did
And work your way up
well im trying it as we speak
Most stuff can just be asked
no responses due to competitiveness or late ones
people make money with these you think they wil give out info that easy?
that's my whole point
Yes
No one makes money out of guides
if you have any specific questions, drop them in #✨│ai-help or make a new topic
There's too much stuff to cover
so do you know someone willing to teach me to stay in touch with the original
To just tell it to one person
my good guy i made several atempts they don't answer
Maybe make a suggestion about more advanced level guides
you might want to look for some audio editing tutorial/courses, since some of them cover your needs
esp here https://www.izotope.com/en/learn/most-recent.html
That too
ok that's good but will it work for the purpose of rvc?
Yes
Plasma can i have some examples of your models?
On like what I use
since you followed the guides
your stuff must be clean as you say
cause mine are not
with these gyudes
That's a dataset issue
you tell me you dont use youtube audio?
like how a doctor needs to know how to diagnose symptoms and apply some cure, you gotta need to learn how to diagnose audio problems through waveform, spectrogram, and your headphones as well as the fixing techniques.
so your philosophy on this is
Ayo? @copper tinsel level 6 !!! 
trial and error alone
i agree but
don't i need a starting point / blueprint to know what im doing?
and reffering to those guides they are not it
well most of them
i'll give you an example
the guides talk about UVR de echo-reverb model
on the chain
but this model i was told that is outdated
and not good
It's not
Outdated at all
There are probably better ones
But it works well for me
so
you said it your self
the guides are not oppted for the best results
just the basic stuff
and that's why im searching for the best and only
Ok but the result varies on a lot of stuff
Not just
The model
I use a lot of audacity as well
Which Is all manual
Work
when you say manual
Thing is
Clean it in general
Parts that the ai can't clean itself
Also there's a lot of testing that each person does depending on their needs
you dont wanna know how mch testing i've done
Like bs roformer the best model for vocals might not be the best
For your needs
Also if from your testing nothing works than you might have a hard to clean dataset
i will give you a song as an example
Cleaning is based on the data it's not the same for every
and i want you to be honest if it is hard to clena or not
Model
cause to my ear its pretty simple
Ok
but apparently the guide's chain wont do it
Produced by & Skive
Mixed & Mastered by Skive @RoughNote Studios
Lyrics in description
Download link mp3: https://www.sendspace.com/file/muzis8
Light on FB: https://www.facebook.com/OFFICIAL.YungLight/
Light on IG: https://www.instagram.com/nerogreco/?hl=en
Skive on FB: https://www.facebook.com/iam.skive
Skive on IG: http://www.thepictaram.c...
also the key objective of making datasets is to extract/do restoration to dry voice audio
hence you need to apply techniques like separation models, dereverb, denoise, etc. though the techniques are quite similar to the one for cleaning podcast audio.
Have tou tried bs roformer
It just came out so I don't think it's in the guides
in reality, recordings might leave little room reverb, not only from the mixing effect
yes
Also there's epoch count batch size
and its good
but i still get errors
maybe theres a different chain i need to follow
What error
that's not written in here
Like it the software or in the model quality
well my models have articulation issues
the voice itself is decent
raspness and some features are well spot on
and yeah i have checked the index usage on parameters
Jokes aside
that's why im askingfor help
im not here to flex
I know
I don't think a master will help a lot more but go for it
do you know where can i find one
but i don't want to buy a model
I don't know about that
and he ignored me when i asked for lessons
so?
Not everyone has free time
And if they have it they might as well make better guides as you say rather that just teach you
they just dont want to spill the sauce dude
how will they nmake money then from comms
ok to sum it up cause i wasted more energy than needed
They still make money even if people know how to do it
The ones on Rvc disconnected
Ayo? @coarse lake level 20 !!! 
i use applio is that bad
Not at all
My pc is shit so I use Rvc disconected
im on paperspace
Oh
If it's the free version
Use Rvc disconnected
If you have the paid version that's good
Not no more I used to make free models
I stopped
Bcuz to many people started asking
And I'm busy as fuck
oh ok
There isn’t lol
I am very willing to teach ya stuff tho
I need some examples first
of your work
https://discord.com/channels/1159260121998827560/1248866741912866826
https://discord.com/channels/1159260121998827560/1246577755215102104
https://discord.com/channels/1159260121998827560/1218969155005317190
https://discord.com/channels/1159260121998827560/1231244519077580870
https://discord.com/channels/1159260121998827560/1236416153731993692
The last one is a pretrain im working on
do i have to make every model 32k? or according to the rx diagram and the void space in between?
cause i see a lot of 32k there
You should listen to the spectrogram
Never use 48k because most people can only hear up to 40k and its just very rare to find true 48k audio
And what is the ideal dataset size I was told anything above 15min is an overkill
Anything above 40 minutes is overkill
10-25 is my ideal dataset size
ofc quality over quantity
is the isolation+ training procedure same for singing / rap / dialogue models?
sorta different
mostly the same
but idk much about singers since ive only done e-girl models
What model on mvsep do you recommend to remove reverb?
uvr-dereverb or foxjoys dereverb
I cant find UVR de reverb on mvsep tho
Also can you accept my frien d request to not spam this forum
you can use deecho-dereverb then
@copper tinsel Yo not sure how relevant this topic is, but heres a few links i gathered overtime
https://rentry.org/RVC-dataset-RX-updated
https://docs.google.com/document/d/1GXyypz0qF6CNvvaV-_7Fn5IiGt56L855HoMdZSfC41w/edit?usp=sharing
In this guide i will be explaining how to use a Paid Software to clean audio for training models.
iZotope RX is known to be the software for denoising audio and the one used by "every" good model maker.
In this guide i'll be recommending mostly free VSTs/Plugins and some paid plugins wh...
RX module settings, what they do and how to use them properly for RVC datasets Spectral de-noise Now Spectral denoising removes noise from signals by analyzing and modifying their frequency components. It identifies noise frequencies, suppresses them, and then transforms the signal back to its...
extracting :)
epic extract method:
mvsep method (make account first and login):
dataset pipeline
instrumentals
crowd removal
before doing allat:
open uvr, get latest version,
go to download center
click on key
enter 02aeb35c203ed0a9
download mdx net instvoc hq2
epic extract method:
get hq flac o...
edit 15.06.24 deton24’s Instrumental and vocal & stems separation & mastering guide (UVR 5 GUI - VR/MDX-Net/MDX23C/Demucs 1-4, BS/Mel-Roformer MVSEP-MDX23-Colab/KaraFan/drumsep/LarsNet/x-minus.pro/Ripple/GSEP/Dango.ai/Audioshake/Music.ai) General reading advice | Discord | Table of content (or ...