#✨│ai-help
1 messages · Page 100 of 1
this is what i keep seeing
Ayo? @wary spoke level 1 !!! 
oh that's outdated
outdated?
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
yeah, been updated in forever
oh-
now that google colab changed sligtly, it broke and won't be fixed
so this is the new one?
you're looking for ilaria RVC to run voice conversion stuff
that would also work but is kind of slow
thank you so much for your help
Hey guys, in the guide it says to revert your model to right before the OT point from TensorBoard. Can anyone explain how to revert a model to a specific step?
Also, is it possible to continue training an already-trained model? If so, how?
Im running rvc 2 locally
why idex file now is so small size ?
Ist which guide? ....2nd yes it is possible to train a model which is already trained. but you need some files such as G_2333333.pth and D_2333333.pth also pre-processd data etc.
which pre-trainer you have used ? if OV2 then it will be small. Also depends on Your Dataset length
are there any alternatives for hina's colab mod?
for index files the case is weird :
if it's small enough, it's proportional to the dataset audio length
It's RVC 2 beta sample length is over 1 hour before it's usual over 500mb for idex file but now idk why it's only 30mb now i iam so confused they update rvc or i do wrong
if it's deemed too big, it's reduced to a fixed size that is much smaller
and the theory goes that this smaller (reduced) index file is diverse enough to behave as expected. But that's the theory
Oh if it's reduce my file size does it effect quality?
I think due to new pre trainer
I wouldn't know much ¯_(ツ)_/¯
Not sure to which degree though
index file stores the accent of your character
Idex file is useful it's correct tongue i usually set it to 0.3
yes. try to change ov2 pre trainer to original
if you want a Big file size 😆
Thank you two senior
@rapid badge does appolio work on your mac locally? It doesnt on mine. Could you please assist me? I sent u a friend request.
Expecting value: line 1 column 1 (char 0) who has the fix to this
Ayo? @fierce zealot level 1 !!! 
training ?
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
runtime?
Use llarias rvc
is bit confusing for beginners. I think
The colab? You press two buttons
check your runtime is connected or not
I know that colab, it doesn't work anymore.
The only tutorial I could find of it is in Spanish 😢
he wants to train
The old easy GUI was not for training
I used to use that exact one
ilaria rvc don't support training
Nvm
I think you have to check runtime. it is connected or not
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
why this ? every time
Sorry, I was gonna use easygui
Ayo? @lime otter level 3 !!! 
I requested a model but I realized that I'm probably gonna have to do it myself
#🤖│bots if you are testing the commands
sorry
you can pay someone to do that
nah
why do you ping me
sorry
mention not.
I've been training this dataset for a VERY long time with the RIN Pretrains. Do they usually take longer than the default RVC pretrains?
I'm guessing that last dip will probably be the best point to stop training right?
Check mel and kl just in case
Wouldn't really recommend stopping just cause g/total got a dip and started increasing... you could stop too soon, so make sure you check the other graphs
Hmm. Okay. Thanks for the advice, this is my first time using the RIN Pretrains
Okay, so I don't know too much about the mel and kl graphs. I'm looking for when they start to trend upwards or level out right? After 750 epochs this is my final result. Judging from the graphs I'd say overtraining begins around the 130k steps mark.
RVC Guides (How to Make AI Cover)
Translation by country
I wanna extract the vocals into the ai vocals
I dont think its ilaria
Im new to this ai song
Oop nvm
Ayo? @rugged crown level 1 !!! 
Wdym? Separate vocals and instrumentals of song? Or use the voice model?
Wlso how can i copy the kink of the modep
So you wanna use a voice nodel, make an ai cover
What model are you trying to use
Melanie martinez
You already found it?
Yep
seems like mel and kl are still going
The onky hard thing now
Is how to fix the error
Like when i am trying toextract the vocals it says error
Can you share me what you found then
So you didnt extract the vocals out of the song yet? Check https://aihubdocs.github.io/en/vocal-isolation--datasets/vocal-isolation/
Last update: Feb 29, 2024
It isn't even public that's why it doesn't work
RVC v2 Disconnected Can I use ov2 after setting it to 32k in Colab? Or can only 40k use ov2?
it seems to also download 32k
meaning usable
wow thx
nevermind
I got a huge problem, whenever I put a voice model from Google Sheets it gave me a JSPEN or file path not found. But when I put a voice model from here, it gave me model not found. How can I fix it? I don’t know if I need to change the codes or something.
Even if loss/g/total goes up and down repeatedly, if loss/d/total continues to go down, isn't it overfitting?
uhhhm, someone can help me? I can't find spanish models :<
I try to use the search but I can't find any
There its a section or something to find english or spanish models?
id suggest checking mel and kl
might be overfitting
Is there any data that tells me what to check for those two things?
Is there any free alternative of Izotope Rx10 denosing?
can you send the links to the models you tried? they may be deleted or privated
Ah, thank you. I found it. If all four things go down, it's good. Then, just because g goes up, doesn't mean it's necessarily overfitting. Even if g goes up, if the other three go down, it's not overfitting, right?
Ayo? @plush compass level 9 !!! 
not really, but its simple
mel is how good it is at replicating the clarity of the dataset, lower better
kl is how different the generated stuff sounds from the dataset, lower is better
Yup - if g goes up it's not necessarily overfitting.
Even if g goes up, if there is no problem after looking at other graphs, it would be right to try learning up to the target epoch. Thank you.
You explain it really simply.
You're welcome :)
I can't get administrator option when i right click on file explorer download
Can anyone help me please
just click the edit button and add your .pth file (index is not needed in okada), like always
what are you using and why would you need admin? what are you trying to install?
a voice model
It says that Im the admin of my system but I still dont have administrator on file exlorer when I right click.
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
Ayo? @woven falcon level 1 !!! 
how do I save my model after the runtime disconnects?
Did I have to save it after each checkpoint?
after disconnecting you can't save it
Hi guys, is there any program that I can use to modify in real time
so you can use it directly on discord for example
thanks 🥲 guess i'll retry again tomorrow
Ayo? @golden moss level 1 !!! 
can someone help me to seperate some persistent backvocals that not being seperated?
I'm doing a training of a model, how it possible that there Is no connection, and the training still going on?
anyone know why my voice keep cutting in and out?
RVC Guides (How to Make AI Cover)
Translation by country
how to cover autotune music?
Ayo? @autumn turret level 1 !!! 
what's the minimum length a database should be to work properly?
(give .index, have distinct voice)
What are you using to separate it? UVR5?
my google colab training got stuck at like epoch 352 for like 20 minutes, anybody know why?
Sometimes it's bugged when you change tabs. Dw just continue ur train
voice changer? If so go to #🔍│help-w-okada
Ooh thank you!
So that errors do not occur, a few seconds (previously very short models like "Minecraft Door" gave errors when generating index, I don't know if this will continue to happen). But for the voice to be distinguished, I would say >2 min
Maybe your environment was disconnected, can you show me?
so i just need 51 seconds of ben diskin talking on a slightly lower pitch
*49 seconds
what colab?
Okkie
is getting close to 2 minutes good enough
how can i make a voice model?
probably only 40% of the voice is now the actual character
https://aihubdocs.github.io/en
Check our documentation! 
Last update: Mar 3, 2024
the part where i run "run-applio.bat" doesnt work
it only says "'env\python.exe' is not recognized as an internal or external command, operable program or batch file."
im following the tutorial where i need to install applio
Ayo? @velvet siren level 1 !!! 
😀 👍 👍 👍 👍
yo is the google collab website not working for anyone else
like no matter what link I put in, it doesnt work
please help
what colab re u using?
i was using the old colab link
the one from the old server
is there a new method?
use the new ones
where
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
guides on:
https://aihubdocs.github.io/en
Last update: Mar 3, 2024
yep
ok cool thanksd for the hellp
ur welcome
Yo there Are AI Cars right on no hesi. But i wanna play it in single player. (Getting bored of thé samé Cars in no hesi. Wanna try Cars that no no there). I tried tutorial but .. the Cars spawn indeed but not on road. On parking lot And their not moving ať all
What the hell?
I swear to god im on diffrent server. But im sure i switched 😭😭🙏😭🙏
Ayo? @zinc mesa level 1 !!! 
they thought this was a different server
Does anyone know where I can find the Raspy/Screaming version of Chester's voice? Did it get removed?
I tried searching for it earlier but I couldn't find it anywhere, but I wanted to make him sing some screams
what is the "label" that RVC uses for training? It's unclear -- does it use words as the labels?
https://www.weights.gg/es/models/cltd4as4t06het1ys92tmbokf
Maybe this model but i'm not sure
Hmm, I'll try this one
I remember there was a model made by bob40 (I think) which explicitly had the words "Raspy/Screaming" on the title
hmmmm i can't found it
its on the c drive
Man why is it so hard to make Chester scream at C4
And transposing it by an octave (12, C5) makes it sound too high-pitched
I've tried three models already 
What's the difference between the quality of the old RMVPE model VS this new updated one?
C:/Users/(My Name)/OneDrive/Documents/New folder (4/My own voice models
moved it to my other documents folder
RVC Guides (How to Make AI Cover)
Translation by country
api-ms-win-core-path-l1-1-0.dll error please ?
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
no thats really old
check the ones on here instead https://aihubdocs.github.io/en/
Last update: Mar 3, 2024
or look above
what are you using, any error codes besides this?
you shouldnt really have a error like that, thats python related
Ayo? @mossy cave level 1 !!! 
Can somebody help me? I have a problem that every time I want to enter a command and I enter this / I only get the integrated ones but not the bots commands and I want to see how I solve this
Are you using a Client?
Or the base version of Discord?
pues la normal bueno uso la versión web no se si cuenta como base
Me too 🫤
The Colab is really outdated
https://aihubdocs.github.io/en/ use the ones from here
Last update: Mar 3, 2024
Like the Applio Colab
this link doesnt work for me on the colab: https://huggingface.co/RandomUser91/newer_and_better_models/resolve/main/DrakeAllDogs.zip?download=true
Can anyone help me, its my first time trying to use a model
To train a model should I have the person say certain words or does them just talking for 30 minutes work
Thank you, I'll read it ♥️
is this a good training script
you can also sing [if you want to make a model that you will use for making covers] or you can record anything in any language
Would this work for covers tho?
Ayo? @south stream level 2 !!! 
if you are just reading or speaking something I think the results will not be very good. I recommend you to sing if you want to make covers [if you are making your own voice model ]
A streamer is having me make one
Whats a good song for them to sing
how to text to voice using the models ?
Ayo? @random turtle level 1 !!! 
do you want to make a model of any streamer
I think ilaria rvc can help you to do this : )
ok I'll try to find about , thanks!
do you want link of ilara rvc colab ?
I would like to use local , is there any tutorial or how to install
what?
Well its a character, they stream on nopixel
do you want to make a model of that character?
yes
thanks I'll check out ! thank you very much!
A person is not sharing there model so I took it upon myself to give it to the community
So he wants me to send him a script to say for training the model
if you have singing data [when streamer is singing something] or you can include normal speaking data, that will work too but you will get poor quality on high pitch
I just need a script
I'm making a Hakos Baelz model and I just want feedback to see if this is a good enough dataset. Also just want to know if I should use RIN_E3 as my pretrain or just use normal pretrain?
if you want a speaking model. you can tell him to speak anything
I want a singing model
yes
lol
Like on god, its safe?
tell him/her to sing some songs. [for better output on high pitch] otherwise speaking will work too
why dont you trust applio?
you can try applio too. it is very good and easy to use with a lot of extra features
Is it overtraining?
Just the gigs, not really dont trust it, just a little suspect
My current rvc shit was barley 5 kbs
and applio is 5,000,000
💀
Did I hit the highest point of training or not?
do g/total not loss/g/total
yes i'll trying to donwload applio because I want to run locally,thanks
Fuck it, im unzipping it
i mean whats 3k and the newest nvidia card if its a bomb?
Right chat?
it not a zip bomb i use appoil
Now?
Should I stop training at this point or continue
use this doc that can help you about over training https://aihubdocs.github.io/en/rvc/resources/epochs-overtraining--tensorboard/#section-4
Okay and
Ima lay down while this downloads
It look like it fine. If the line reaches its lowest point and starts going up then it is being over-trained. This picture look like the model is not overtrained
So I can keep on training it
sure just make sure it not being over trained
Nice
I know already it's not my first time
But I never knew about overtraining
that why this server provides a doc about model making
Yes
On the google collab, the second part where I add the link and run it, there is an error. "JSON decode error" any idea?
how did you fix it?
can you help me ? in applio what folder to put the index and the pth from the models I can't find
I read the rules where I get stuck in step 2/3, so I can’t write in "voice model". If someone has a solution please
What are the best settings for making an RVC model on Colab?
Don't quote me on this but very, very clean datasets uses RIN_3 or Mangio crepe. OV2 is used for 6 minutes or less
stick with the standard pretrain
Well in your opinion what provides results
Ayo? @rugged stirrup level 9 !!! 
Use a 32k model. It will sound better than 40k with less robotic artifacts. But make sure to convert your wav file to 32,000 sample rate
I thought it automatically does that and I just feed it the MP3
One of the problems I had was keeping it at 48k as a default and it's trying to look for the missing frequencies. I think that's how it works?
Ohh I thought 48k was the quality of the voice like the sample quality
Like how audio at a lower sample rate sounds worse or lower quality
You want it to match so if you have a 40k dataset for sure, then do 44,100 or 40,000
I'm not sure audacity is true resampling since I do it on Izotope
I think I always have a 44.1K
RVC Guides (How to Make AI Cover)
Translation by country
What should I use for this? or do i need to down sample it to 40K
Idk how to check that
I have 20k on mines, but training on 40k sounds weird for me. No always use a wav file or flac
I would redownload your dataset in wav instead of mp3
Oh ok
40K/48K tends to have strange pronunciations with words
Yeah, it didn't sound like my model so I scrapped it.
I mean 48K/40K does sound like it but when it gets to the high pitches or something like that
How do I export multiple files at once cause I have a split dataset
Export range > multiple files?
Alr thanks
anyone here use mangio? i keep getting an error when i try to convert after a few seconds
ive tried google colab, rvc gui, and now mangio isnt working for me too
help??
why use the help channels out of all places...
If you are using mangio colab, it won't work atm
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
Ayo? @rugged crown level 2 !!! 
@rugged crown Bida ang saya
Help
Can you screenshot the console
Help pls
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-5-a8e5665565a0> in <cell line: 4>()
2 #@markdown Link the URL path to the model (Mega, Drive, etc.) and start the code
3
----> 4 from mega import Mega
5 import os
6 import shutil
ModuleNotFoundError: No module named 'mega'
NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.
To view examples of installing some common dependencies, click the
"Open Examples" button below.
What colab is it?
Snippets: Importing libraries
No the link to it
No the link to this
hey how t use voice changer on discord
You need virtual audio calble
Link to the model: https://huggingface.co/PlayerBPlaytime/My-Models/resolve/main/Melodie.zip
I hope that's right
Ayo? @vocal hedge level 1 !!! 
Its soon?
it say it cant detect my voice
Ayo? @manic sapphire level 2 !!! 
@manic sapphireCheck if you're using the right input device
wait
If you're using the voice changer on Discord, you need to set up the Virtual Audio Cable as an input on there (and make sure to have it as an output on the voice changer)
ooh wait
You can use your headphones/speakers as a "monitor" output on the voice changer (if you're using Okada) to hear yourself
Don't use it if you're in voice chat tho (if you're using speakers)
Okada takes your real microphone audio as an input, transforms it and outputs it through the virtual audio cable, if you use your mic on Discord as input people will just hear your mic audio, if you use the virtual audio cable while you have the voice changer on they will hear that
Output is still your headphones or speakers
On Discord
oo
So you can hear everyone else
where do you downlaod the software that you will use to voice change?
Ayo? @buoyant hearth level 1 !!! 
Idk, you can find more info on #🔍│help-w-okada , pinned posts as well
k
There are different versions
Depending on if you have Mac or Windows, and if you have a Nvidia or AMD/Intel GPU
Cuda is for Nvidia, DirectML is for AMD/Intel
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
and for wokada ask help in #🔍│help-w-okada or #1192011222023950368
so, this one right?
aight, thx man
You have a Nvidia GPU?
yea
Then yes
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
why when i talk and the voice change have long delay
could be your gpu or quite literally anything
if you're on amd and using a downloaded model, check this:
https://rentry.co/VoiceChangerGuide#uploading-models-amdintel-gpus-and-fixing-its-laggy-issue
Other Links
Antasma's Local Error Fixes
Antasma's Colab guide
Sushi's useful Links - You need to follow this if you are on an AMD or INTEL ARC graphics card
Frequently Asked Questions
Jump To The Important Sections
Other Links
Download NVIDIA
Download AMD/INTEL GPUs
Download Mac
Online Hosted Alt...
Can anyone help me understand Tensorboard a little better or give a link to an indepth guide? From what I understand I'm looking for the low point in the loss/g/total graph before it starts to increase. What do I look for in the loss/g/mel and the loss/g/kl graphs? And which graph is more important? Here is the tensorboard of the current model I'm working on. Can someone tell me at which point they think overtraining being and why?
Please help
RVC Guides (How to Make AI Cover)
Translation by country
Can you show us what errors are output in Colab?
be sure you manually selected everything before clicking convert
ok
You didn't select the model
you put a drive link
You need to download model and refresh and select it
anyone here use the replay app?
Ayo? @near dagger level 1 !!! 
Nope.
But if you can't use it (likely due to PC Specs) you can use Ilaria RVC instead
Or a Huggingface one
Now you must drag an acapella audio of yourself or someone singing where it says "drop file here"
Or press on "click to upload" and look for your file in your file explorer and select it (For phone)
And that's it
Oh sorry i was saying that he put a drive link in the model selection sorry
You can put or a Drive or a HF one.
HF ones are less likely to get taken down
Hell, you can even use huggingface to store your personal files and put them in a private repository.
And it would be almost the same as paying for TBs of space in OneDrive or Dropbox but FREE
Yes true 😄 i was saying to him that he put the drive link in the model selector instead of the download section
Like, he put the drive link in the model name and not in the downloader
So he didn't download
Good 😄 and you Leo?
You

I'm doing pretty fine
I've recently moved to another apartment.
Oh nice, i hope that is all ok! 😄
I wasn't pretty active due to being occupied loading boxes and due to not having an internet connection on my new home
Thanks for your nice wishes.
Is ok! Don't worry, in last period i was a bit busy too 😄

Hehe i hope you're doing very okay
Thank you so much 😄
sorry to bother but uhh is this normal?
it just keeps going up
Are you using Huggingface spaces?
i tried the ilaria rvc u suggested
dang got an error
what do i am very new to this i just wanna hear sily character sing silly song 😭
Don't use HF spaces
Use this colab instead:
sorry for all the questions but how exactly do i use this? 😭 is there a tutorial somewhere?
Ilaria got a tutorial but it's on italian
So i'll summarize the way how you can use it
First of all, run the first cell
omg thanks much appreciated
Run the first cell
Then wait till it finished installing
Then run the second cell and wait till a "Public URL" pops out and then click on it
yes .. Okada
where i can dowloand this
got it or not ?
-rt
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
ok thanks
xocas
Yup me too
I contacted the dev already, dw
Should be fixed later today
yw. In the meantime use the Applio Colab or RVC Disconnected
https://aihubdocs.github.io/en/ here are the guides =3
Last update: Mar 3, 2024
i probably will
hey, so i've been doing everything correctly, I'm using the colab version and then hit the Install step, then grabbed the link for the villager news from element animation ai model (https://huggingface.co/SyberGen/Minecraft-VModels/resolve/main/VillagerElementAnimation.zip) and copy pasted it into the 2nd Step when not even a second passed that it immediatly stopped, heres what it shows me afterwards... if anyone wants to help me out i would really appreciate it.
JSONDecodeError Traceback (most recent call last)
<ipython-input-13-40c3d8edfd86> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
――――――――――― ↕ 5 frames ―――――――――――
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
U re using and outdated colab, try the new ones:
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
Oh...
u can find guides here:
https://aihubdocs.github.io/en
Last update: Mar 3, 2024
thnaks
Ayo? @lone cradle level 1 !!! 
https://huggingface.co/spaces/IAHispano/Applio Hello, does anyone knows how to use this version of Applio?
do you want to make a cover?
do you have any voice model?
yeah, on my files, on my pc
okay.. go to download section and drag,drop pth and index one by one
or if you have it on hugging-face just paste the download link of hugging-face
how?
While installing Applio , I'm getting this error. How can I fix that ? **Applio-3.0.8\app.py", line 1, in <module>
import gradio as gr
ModuleNotFoundError: No module named 'gradio'
**
have you uploaded your model on hugging-face?
@sand iris
on the download section?
like
Ayo? @brittle wing level 1 !!! 
i think that's how it works on huggin-face
how'd you install applio ?
??
i'm doin' it rn
you can simply drag and drop your index and pth files here
Been a long while since I ran applio, but from what I saw the scripts are broken
have you tried using the release archives instead ?
just download the model from any source (hugging-face, weights.gg etc) and then extract it and after that drag and drop pth and index
Where can i download archive release ? From github ?
Ok thanks
@sand iris I have one more question. Sometimes UVR doesn't use my GPU. I don't know the reason , i have rtx 4050 mobile laptop
I don't know about UVR
I've got this crackling sound when I speak in Discord but not in the VoiceChanger itself. Anyone know how to fix?
ask in #1192011222023950368 or #🔍│help-w-okada please
Ah sorry
Ayo? @old shuttle level 1 !!! 
be sure u check the gpu conversion box
It's already checked. I think in some GPU's like rtx 4050 , MDX-net cannot be used . I can use VR arch with gpu
i need to train my own voice canb somboedy help me with that
try to update uvr then
Last update: Mar 3, 2024
you can do this by clicking the bottom left icon where you download the models, and click check for updates
anyone know why this isnt working 😭
i was using a different model just a while ago and it was fine
i also tried uploading it to gdrive making sure its accessible to everyone but its still the same error
check if that model has both the .INDEX & .PTH
if they do, check if they are the correct ones:
https://aihubdocs.github.io/en/essentials/voice-models/#they-are-made-up-of-two-files
my rvc real-time struggle alot when I use intensive CPU task. I though this is using only GPU?
Ayo? @sacred lichen level 1 !!! 
can someone help me
for real-time voice changer ask in #🔍│help-w-okada
RVC Guides (How to Make AI Cover)
Translation by country
Do large datasets need to be trained for 300 epochs or could I train it on like 80 or 100 I have about 3 hours of audio
there isn't a way to know the right amount of epochs beforehand
Oh
https://aihubdocs.github.io/en/rvc/resources/epochs-overtraining--tensorboard/ you can learn here =3
Last update: Feb 10, 2024
so use TensorBoard instead
How do i get tensorboard
it's explained in the doc. please read thoroughly before asking questions
🙏
Cause some models sound robotic for higher pitches and some that are lower epochs sound more natural is that a sign of overtraining
it depends. yeah sounding off can be a sign, but unclean samples when inferring can also cause that issue
I try to keep datasets as clean as possible using UVR and stuff
it could also be that the dataset didn't include high notes. so when you try to make that model sing a high notes, it "glitches" up like that
Ohh
besides cleaning, variation is also important for datasets
https://aihubdocs.github.io/en/vocal-isolation--datasets/datasets/ you can learn more here =3
Last update: Jan 14, 2024
Oh ok
When you make a dataset, the .wav should be around 10 seconds long, does it matter where they end? (if it cuts out at silence or if it cuts out during middle of a word)
Doesn't need to be cut like that
RVC automatically does that (in the preprocessing step)
Am I in trouble with how jagged the curve is?
smoothness of 0.987
also that finiky end part was because I accidentally stopped and then resumed my trainning
probably not
are you on a low batch size?
Have not tested yet, as I have quizzes
I'm asking because when I let RVC do it in the preprocessing there's many super short ones (sub 4 seconds long) and in a bunch of them it kind of cuts off in the middle of a word
just let it preprocess
alright then
The 137.0.0.1:1888/ isn’t opening
How I can fix "File "/usr/local/lib/python3.10/dist-packages/transformers/utils/hub.py", line 96, in <module>
PYTORCH_PRETRAINED_BERT_CACHE = os.getenv("PYTORCH_PRETRAINED_BERT_CACHE", constants.HF_HUB_CACHE)
AttributeError: module 'huggingface_hub.constants' has no attribute 'HF_HUB_CACHE'" this error
what were you trying to do
I'm just run Step 4 after I ran this error is show
Ayo? @thick dagger level 1 !!! 
is 660 epoches good?
which colab
is there a way to disable backup vocals on AIcoverGen even at -20db its hearable ? or what tool do I use if I just want to infer my model, and no extra processing ?
Ayo? @hoary summit level 1 !!! 
outdated colab
use rvc disconnected
Do you have a new one ?
Ayo? @raven burrow level 1 !!! 
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
Thanks bro
Do you have a guide ?
Thx bro
all g
guys I don't have the requirements for local rvc and I am not using colab because paid 😭
any solutions?
just spent hours making a dataset aswell
holy this server is dead
colab is free
it says I need premium or something
which colab
-colab
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
rvc disconnectred
Can someone help me with waves harmony?
yup it has edge tts mixed to it
Today around 10am I opened the collab but when using the clone and install option I got a warning that said "due to the limitations of collab (and something about the gpu)" so I started the code To make the cover he told me that I needed an NVIDIA graphics card, I don't know what that means but I need an answer
Ayo? @mortal plinth level 1 !!! 
XDD
(I speak Spanish too)
Porque es mi idioma xd
a
It means you finished your free daily gpu of colab for using it
either use another google account or just wait for tmr
Guys
I need help
JSONDecodeError Traceback (most recent call last)
<ipython-input-6-1d7b23896ae4> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
Ayo? @still stump level 1 !!! 
Google collabs keeps giving this message whenever i try to download a model
Even the default Kanye model
please use a newer colab
this is an old one
Idk where to find a New one
-colab @still stump
Suggestions for @still stump
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
for inference applio or ila rvc is good
normal one should be workin
?
yes
Cool
That’s the exact same error I got whenever I tried to download a model
please use a newer colab
-colab @mossy cave
Suggestions for @mossy cave
- Applio, by IA Hispano Google Colab
- AICoverGen-WebUI, modded by Hina Google Colab
- AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
- AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
- Ilaria RVC, by thestingerx Google Colab
- RVC Disconnected, by Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- Applio, by IA Hispano Huggingface Spaces
- Ilaria RVC, by thestingerx Hugginface Spaces
- RVC-HFv2, by r3gm Huggingface Spaces
- AICoverGen, by r3gm Huggingface Spaces
- Advanced RVC Inference, by r3gm Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
applio and ila rvc are good for inference
Question: can I use Google Sheets for models?
what?
oh
if its a working huggingface link
better to look in #1175430844685484042
is there any colab working?
hello
Ayo? @soft night level 1 !!! 
I'll try it again today and see if i still have the issue
hey
im doing rvc disconnect
what is ov2
pretrain tyoe
and what should i choose for it?
RVC Guides (How to Make AI Cover)
Translation by country
you can use ov2 if you have a very short dataset, if you want
i woudnt use it for anything over 6min
some say 8
me personally i rarely use it, its really good for very small datasets tho
please help
when i try to use some of the voice packs it says
pth file path cannot contain Chinese characters
how can i fix this
please help i follwed all the rules and cable audio drive and it still not picking up sound
well, dont have chinese letters then
make sure your file path is not too long and doesnt have any spaces in it
or special letters
can you send a screenshot of your settings?
ohhh
Ayo? @sharp mesa level 1 !!! 
it thinks latin is chinese😅
im sorry but i have a new problem
it says AttributeError: 'RVC' object has no attribute 'tgt_sr'
now
what are you trying to do?
can you show the full error?
probably asked a bunch but i couldnt find an immediate answer
whats the diff betwen RVC 2 and 1?
Hi
I used it for a 1.2 min dataset
went 500 epochs no overtraining
what files do i use from this now
just use v2, its newer and better tech basically
not sure if what im using is v2 lol i downloaded MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a today
yeah dont worry about it then
doesnt matter when you are using a voicechanger, just make sure to actually use rvc models
Other Links
Antasma's Local Error Fixes
Antasma's Colab guide
Sushi's useful Links - You need to follow this if you are on an AMD or INTEL ARC graphics card
Frequently Asked Questions
Jump To The Important Sections
Other Links
Download NVIDIA
Download AMD/INTEL GPUs
Download Mac
Online Hosted Alt...
and follow this guide 👍
anyone use waves harmony?
and if so I can I make the harmony with the ai?
since you used colab, you can try out ilaria rvc https://aihubdocs.github.io/en/rvc/cloud/ilaria-rvc/
Last update: Feb 25, 2024
or some of the other options
thanks
dont think anybody here does
maybe @trail iris
the waves plugins?
yes
yeah
you can use it ofc
but then it will just end up getting more robotic
but u gotta make sure ur model is like
clean
how do i upload a picture here lol
appreciate the help
you should get perms after like 3 messages 💀
whats your issue?
oh all my old msgs are gone, im assuming the server had a purge at some point??
anyway im trying to train rn, im only at like 2 epoch but its going so slow and the training graph is just flat
its rvcui/tensorboard
Ayo? @strange quest level 1 !!! 
gg i got picture perms
this is what i used 
does more epoches mean higher quality/accuracy?
Ayo? @foggy heron level 1 !!! 
not always
every model is unique and overtrains at different points
what is going on here 
main factor is the quality and lenght of your dataset
are you training locally or using a colab?
local
ah ic thx again
and how big is your dataset
like 20 mins
lol mine was 1 minute
whats your gpu?
trained it to 500 epoches
3050, which i heard wasnt good for it 
yeah its the graph thats throwing me

ohhhh
now it looks like a graph 
whats the predicted time for the dataset around?
yuh
it says heelllppp heelllppp pleaassee 💔
and what batch size did you use
ita taken about 40 mins ish to get to 3 epoch
thats way to much wtf
IM SAYING!!
buh
you will lose progress but you didnt get far anyways 
i had a bug where it took longer to train, because my cmd window wasnt focused lol
click on it and press enter once
the way nothing happened
Ayo? @strange quest level 2 !!! 
i would restart and make sure your settings are good
you cant continue training like this
if you are using mainline it should take 30-40s max with your specs
😭
ok so reboot my whole pc and try again... 
send a screenshot of your settings before you start training again
on tensor?
yeah restart your pc immediately and start from scratch
also are you running low on space?
is it working?
i restarted, just setting it all back up now
you restart your pc right?
yas
and that's how you got the colab working again
you cut off the part above
who where how
the tut said using tensor u put it to like 2000 
Ayo? @strange quest level 3 !!! 
thats for colab
also where is 32!! its just not there for me
are you on the latest version?
there a problem
so uhm so like
at this point idk anymore LMAO
should i just ignore this?
you are almost there 😭
just change the stuff i mentioned and if you are impatient ig you can use 40k
gotta go
thank you... idk if im on the latest version so we'll just run with this lol
so I just installed the rvc
and its a blank screen
and on the cmd prompt it just keeps saying "Wait web server"
cuz my doing what hes doing but sometimes I hear the pitch going carzy
Are you using the real-time voice changer?
i was using the first one
yes
but i may just use the 2nd one instead and see if that one works
what am i doing wrong
Uhh iunno what you mean
Nothing. EasyGUI is bugged
It should be fixed by tomorrow
is there any other gui
alr
I contacted the dev this morning but he hasn't responded yet 
dang
Yeah for now you can use RVC Disconnected or the Applio Colab.
Check the guides here
um sorry was watching a yt video and 1 said one was the first one and one was the 2nd one so basically the one that was not working was called MMVCServerSIO
Ayo? @brittle wing level 1 !!! 
Ah I see. For Voice Changer stuff use #🔍│help-w-okada
alr ty
I installed both Im pretty sure whats the difference between the 2? do they both work for real time voicechanging or is 1 only used for music and the other is used for real time voice changing
No clue. I don't use the Voice Changer
So ask in the other channel
alr ty for directing me to the right channel 🙂
im getting all of this now and tensor is not moving 
Are you following the docs?
yas 💔
Which RVC are you using?
i just downloaded this
https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/RVC1006Nvidia.7z
Has this updated yet?
yea it has a tiny line now lol
but its weird. the tut says to start tensor before i train and select the model but mine doesnt appear until i start training
if that makes sense...
It only detects models that are being trained
Oh no that's not good
i thought so but the tut confused me. made it seem like it was supposed to show before u start lol
not the irl name usage
i think my spotify is on my profile anyway which has my real name lmao
Oh yeah I just realized. But I don't think it's supposed to look like that either
in loss
@strange quest
help what does this mean? 😭
Ayo? @near dagger level 2 !!! 
Crap another Colab error 
Try rebooting the runtime
Click the downward triangle in the upper right corner
i do
Can you send a pic of g/total?
it wont let me find it lol
which colab

Alr seems to be doing fine
How many epochs so far?
well idk whats going on but it says we're on 23 epoch 
For how long have you been training?
Wha
omg thanks! i think it worked!
yw mx ^-^ have fun =3
batch size per gpu?
Yes
i did 8
@little urchin ask here
i dont know if this is the correct channel tho
rmvpe_gpu
What GPU do you have?
Also why TETSKI?
3050
Ayo? @strange quest level 4 !!! 
Crap then what the heck is the problem 
RIGHT
How much VRAM is there available?
If I was you I would restart the puter & train from scratch
where can i download real time voice changer
thank you


