#✨│ai-help
1 messages · Page 8 of 1
So I need to just go with whatever the lowest value is?
The lowest value on my graph has really low epcohs tho
use higher epoch ones then
test em out
lowest point might not be the best version of your model
How so
the best version of ur model is a balance between overtraining and undertraining
where it isnt overtraining but it also isnt undertrained
gotta test tho
So then the value between the lowest and highest should be kinda the optimal one
not exactly
just test out stuff
Also I don't have the steps that line with the epochs
@astral perch has reached level 7.
doesnt weights tell you the steps
like, the pths have the steps in them
It does but the lowest that I wanna test has 18.600 as the steps but in my weights folder the closest to it has 18.560
And another one 18.640
Then use that
The one higher or lower ?
Try the one higher
Np, glad to help
hi,, rvc has been completely fine until today,, the delay keeps going up to 25-30 seconds the second i turn on the vc
GPU set to 0?
-rvc
RVC Guides (How to Make AI Cover)
🇺🇸 English Documentation
Credits: AI HUB France
🇺🇸 English FAQ
Credits: AI HUB France
🍏 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
- 🇫🇷 French Documentation
- 🇮🇹 Italian Guides or
-it_g - 🇷🇺 Russian FAQ for RVC/Voice changer
- 🇧🇷 RVC Guides in PT-BR (
-rvc_br)
how do i do that
@nimble birch has reached level 1.
Wait, RVC or real-time with Okada
real time
More things you have open, slower it is
it can’t be that, it’s worked fine with games open before and now it’s the only thing open with the delay constantly increasing
Can I see a ss of the client?
sure shall i sent here or dms
How do you open up an ai once It's downloaded?
@jaunty skiff has reached level 1.
What do you mean by opening up an AI?
logs/manualtraining or logs/finished
trying to use the rvc gui for colab but it seems to be taking a while to upload a zip, is there a way to download straight from weights.gg?
I requested an ai, and I downloaded it through the dms, but I don't know how to open it.
using RVC
you can use mangio or applio
Wordband which is a discord bot also works
Oh you need a fork then
Try the Colabs and read the guides so you can understand
How do I download Mangio or applio?
how do i change it
@south vine has reached level 1.
How do I download all this for Mangio?
im just now getting this error too. any luck finding a solution?
Rvc training does work with training other languages right?
Nop. My best solution was to put like this N:\Applio\datasets\FolderName\Voice.wav
both the .7z files are basically zip files which has everything so you just download it and then you unzip it to get the folder with all the mangio crepe stuff. the infer.7z only has infer/making song covers while the infer_train.7z has infer and training so you can also train your own models, although u need nvidia gpu to do it
I can't preprocess data for some reason
covergen says my file is not a zip file when, well, it clearly is
anyone knows how to fix that?
Process the Dataset, You can't do Feature Extraction without it.
What's the error?
runtime\python.exe trainset_preprocess_pipeline_print.py C:\Users\Jorgi\Desktop\TOY 40000 8 C:\Users\Jorgi\Downloads\AI STUFF\RVC BETA v2 0528\RVC-beta-v2-0528/logs/ToySFM False
Traceback (most recent call last):
File "C:\Users\Jorgi\Downloads\AI STUFF\RVC BETA v2 0528\RVC-beta-v2-0528\trainset_preprocess_pipeline_print.py", line 20, in <module>
f = open("%s/preprocess.log" % exp_dir, "a+")
FileNotFoundError: [Errno 2] No such file or directory: 'C:\Users\Jorgi\Downloads\AI/preprocess.log'
No space in the path to the RVC folder.
It's not recommended to have any spaces or special characters.
Those may cause issues.
What folder? "AI STUFF"?
@manic ravine has reached level 1.
ohhh i think i was putting my audio in the wrong folder
it goes in the weights folder right?
What does this mean?
I let it always on 8 but for now i wanted to ask what that means 😄
'AI STUFF' and 'RVC BETA v2 0528' have spaces.
Rename them to 'AI_STUFF' and 'RVC_BETA_v2 0528' for example.
datasets folder. Inside a new folder there.
Can you help me
@hardy hawk has reached level 1.
That was it, thank you!
thank you. cant believe i missed that folder 😭 i kept trying to make different dataset folders
@red mesa has reached level 1.
You place all audio (WAV or FLAC) for your model in datasets/ModelName
What's the cause of this problem with Local RVC, and how can it be solved?
No zip file for the dataset.
For you it says 14 on the Top right
For me 8
Do you know what this one means wait
Don't use a zip for the dataset. But a normal folder.
A normal folder with the wavs inside of them?
So download these two?
Yes.
Also, don't have spaces in the path to RVC.
Not sure if RVC works on OnneDrive.
no infer 7z has only infer option while infer train has both infer AND training so it's like if you don't have an nvidia gpu or you don't want to train models and only make song covers, only get the infer.7z not the infer_train.7z
but if u also want the option to train just cause then get the infer_train.7z
You need one, not both.
Ok, the top one or bottom?
Do you wish to train your own models?
Oh I don't know how to do stuff like that at all XD
what does the highlighted text mean
no idea but i do get that error too once the training stops
or fnishes
but seems to not have a difference from i can see
no idea
Well, if you plan on training you will need INFER_TRAIN.
If you're not going to train models, you can go with INFER.
Can anyone explain me this?
i think thats just how many cpu cores you want colab to use
max is 2 on colab
but on local it depends on your system
No its Local
I let it at 8
Before i brick something because idk what Number i can put there
Does anyone know how to solve this error?
@dapper barn Has this error ever occurred for you?
@olive monolith has reached level 2.
Ok, I have it all extracted, what do I open to activate the modal?
Launch go-web.bat to start Mangio.
Where should it be in the folder?
@jaunty skiff has reached level 2.
Inside the Mangio-RVC folder.
I can hardly read that but I think the 11th file is 'go-web'
yes yes it is
What batch size should I use?
One that's a clean division of the number of audio clips in your dataset
How do I put a downloaded model into Mangio?
ModelName.pth goes in weights.
added_IVF*.index goes in logs/ModelName.
it should be one of the tabs there
Which one is it?
Oh ok. I'll keep trying
bro why whenever i tried to load in a model on hugging face it says
theres been an error
What error
When i lauch https_start the rvc client doesnt open it only breifly open console then closes pls help i have spent hours
put the .index file in \logs\GordonRamsey and .pth file in \weights
for any other models you want to add, it'd better to make sure the modelname.pth doesn't have spaces, underscores, and possibly some special characters (except some like '-' is fine) so the "Auto detect index path" can read its index file
Is that with Mangio?
yes, though I also modded applio a little to also work like that
Where do I go to find the /Weights?
Hi
How do I resume training on rvcDisconnected?
I have the processed data on my drive and saved the latest epoch
Ok, I got it, I'm trying to request another ai model, which is a Kidaroo VoiceForge ai, but when I try to request it with free reqest, it won't let me
is there a way to stop RVC from outputting WAV files for voice conversions
you mean another output file format? the latest Applio can do so
im talking about mangio RVC
in mangio there's Export file format but for batch conversion
is there like a configuration file in mangio for the normal one
idk, you may need to mod the mangio itself, especially infer-web.py in the mangio folder and probably something else, or simply use Applio instead
yes
how to do that, is there a tutorial or you can explain?
Just stop at the lowest place, like on the second graph the lowest is about 12k
If i change the model at hugging face, does the model on wordband change too, or does it stay as the old model?
what browser works with tensorboard on colab again?
Firefox
tnx
aight imma use applio gui with it next time
I kinda don't recommend the Applio Gui Colab IMO for the fear of banning my Google Account.
I use the rvc v2 Disconnect
im just using it atm for my own model test using an alt so it should be fine
OK! I just have one Google Account really
@tough crane has reached level 2.
i have like 7
but most of them are not throwable alt accs so ye...
One message removed from a suspended account.
You change the runtime type
On the top right corner, there's a dropdown button, click it, and see which one has the option to select GPU
One message removed from a suspended account.
Wait for a few hours, try again, etc
One message removed from a suspended account.
If i change the model at hugging face, does the model on wordband change too, or does it stay as the old model?
Hi, i tried to process dataset but i ended with this error... I downloaded Mangio RVC TRAIN and run via go-web.bat
FileNotFoundError: [Errno 2] No such file or directory: 'D:\RVC\Mangio/preprocess.log'
nvm i found a solution
@serene hound has reached level 1.
i just waiting for solution
you need to upload a model first
then refresh that part of it
thanks to wanna help, what should I do to add a model?
there should be an upload model tab there
my logs file like this, when i tried to run tensorboard, it just said that there arent any data
lemme run my verision. applio but it should still have the same stuff
where? can you show me?
im booting up mine.. its on colab so its gonna take while
okay! sensei 👍
so, i have upload those on "weights"?
the pth
i know but that both files must be placed in "weights" right?
you have the gui so why?
does it have this part in the Resources tab?
i don't have it 😕
can you screenshot the tabs?
ah so only download
you can use the link on hugging face for the model
the download link
where?
@brittle wing has reached level 6.
from hugging face if the model is in hugging face
sorry but show me the screenshot to be clearer
the zip file on hugging face
I know what it means, maybe I can try
it needs to be a zip file...
i gonna do it but first to make sure
where can i download onnx models most of them are pth models there.
So if it has been uploaded, both files will be automatically detected and will be there?
or via zip?
it needs to be a zip before you upload
i see, sensei
can anyone answer me
if its the same name then it will also change
since it just downloads/redirects to the original hugging face
unless you change the link in your post.. that wont update
ty! i was worry that i have i register another model if i have a new one
wait, By the way, where did you put it?
you copy the link from the small download button
then put it on the download model part of the up
should look like this
so, after i upload it on my files, i gotta to do that?
yup
for some reason when i try to continue training on RVC Disconnected it doesn't continue from where i left off
and wait what
why did it not save
Is there a way to create AI covers online without downloading any software?
@main patio has reached level 2.
technically you can do it on discord with the wordband bot
just do /cover
Where can I do that?
it totally works! i can finally use someone voice in my TTS, extremely thank you very much!
guys where can i find the timestamp of the trainning in my logs
What ai are you using?
some Puella Magi Madoka Magica's Voice Model
Does anyone know how to fix this applio error?
2023-10-18 15:55:42 | ERROR | root | Error on reading from the event loop self pipe
loop: <ProactorEventLoop running=True closed=False debug=False>
Traceback (most recent call last):
File "asyncio\proactor_events.py", line 769, in _loop_self_reading
File "asyncio\windows_events.py", line 817, in _poll
File "asyncio\windows_events.py", line 462, in finish_recv
OSError: [WinError 121] The semaphore timeout period has expired
mid training
I mean what AI are you using?
rvc
RVC HF
Where can I use RVC?
welp
btw this for voice cloning, right? if yes, why there's a error?
Your runtime has been disconnected due to executing code that is disallowed in our free of charge tier. Colab subsidizes millions of users and prioritizes interactive programming sessions while disallowing some types of usage as outlined in the FAQ. If you believe this message is in error, file an appeal. Please include any relevant context about your usage.
Can this be resolved?
It happens in AICoverGen - NoWebUI, by Ardha
i have a same issue, you better use this but this gonna confuse you : https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=ZodNcumpg-JM
use the one w the web ui it works idk why
this isnt for inference only for training
you mean this?
https://huggingface.co/spaces/r3gm/AICoverGen
ahh, sorry, I didn't see all your posts
i was once a qc here but got banned for sum stupid i did if any mod sees this message
It is asking for Sign-in, and when I click on it, it opens a new tab and closes, and nothing happens
@magic hill has reached level 1.
What's GPU?
just use your google account
It is basically showing this:
Your Google Account was signed out on a different tab. Sign in to continue
It is not letting me to...
hold on
Having trouble getting tensorboard to work locally, i've got python 3.10 installed, pip installed. but pip commands aren't working python when trying manual install
what do i do
I'm trying to see where the model overtrains, im retraining one of my models to be higher epoch and then going to reupload it
10
12 gb allows up to batch size 16, but I'd recommend around 4-8 regardless VRAM. can try different values till you find best model epoch as possible.
6 it is!
On a 1080ti is it safe to leave at default ?
I’ve already got my model training
Waiting hours for it to complete
NOTE: There is no such a thing as perfect batch for every case. Each voice is unique, the dataset is unique and one might work well with 4, some might like...
sooo...for a 10 mins batch of audio I?
for 10 mins I think you can try on 4, then can try different, maybe 6 till you find which is the best for your model dataset
ty
ModuleNotFoundError: No module named 'onnxruntime'
Press any key to continue . . .```
Can someone help, please?
whats diff between G** and D**
pip install onnxruntime
Thank you.
what do i do
u have a RVC 1 or RVC 2 model?
RVC 2 model
send model link
where did you get this link?
what did u use to train?
RVC disconnect
isn't that RVC 1
...you serious?
shrug
i put RVC 2
it has disconnect in the name
b-but i put this
@stark grove has reached level 3.
doesn't count?
so wait i have to do the entire model again?
which should i use?
what did u use for training
rvc disconnect
colab? not colab
colab
send link
shru
g
I can't help u, I use local
damn
Noob question, what are index files for?
even though I choose rvc or gpu, it uses too much cpu and the audio endings are cut off. can anyone help about this? ^^
-colab
Google Colabs
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
Wrong folder - check the weights folder
Else you screwed up and didnt export the model
That's a training file if you didn't know
oh
They make the result more realistic by applying accents and stuff like that
What F0 are you using? Dio and Harvest utilize CPU
Thanks, I was missing an index for a model and used one from another model. It worked fine, so I was curious.
i need serious help, i traineed two diffrent models with two diffrent datasets and they sound the same and not simillar to real dataset? i do something wrong?
every dataset converted and trained sound the same no matter if it's female or male
high or low voice
it sounds like some "default voice"?
idk
you're probably screwing up on the dataset part
maybe loading up a dataset you didn't want to? ifk
what do i need to make dataset? i tried to make folder, name it and put on this folder some audio files
is it right?
or i fricked it and that was the answer why it's not work
yeah, then you gotta put that path correctly in the process dataset step
not the files themselves but just the folders path
like that?
How to continue training?
Yeah
Set the same settings (no need to process dataset or feature extract again), change total epochs and/or saving frequency, then start the training.
i think i did it before
Hmmm, I'm not sure what you're doing wrong then
Normally this isn't what occurs unless you're forgetting something
But I don't know exactly what
did you have nan loss values during training?
i think i saw something like that
@serene hound has reached level 2.
hey, ive been having this issue as of late
Before i could to anything with the program and it'd be completely fine, but now every time i use it while i have a game booted up it starts to frequently skip or repeat itself
i don't know why or how to fix it
...gaming isn't gonna be good with Voice Changer
unless the game doesn't use a lot of the GPU
then I guess you may need to update the Nvidia driver, or roll back to known good something like 531.79
i'll try
Ok
@near palm has reached level 1.
i got some update
i will try to install, and then i try to make something small just for testing
need to be studio version or game ready would work?
installed
game ready for most purposes
umm
i accidently deleted whole message...
damn....
everything here is ok, right?
i need to process data first
then feature extraction
train feature extraction
and train model
right?
process data
log from feature extraction
feature index works as well
idk what is this but i hope it's ok
INFO:Seba:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=nan, loss_kl=8.723
hmmm
when it ends
hmm
in the folder of mangio rvc i have these bat files:
compile.bat
go-realtime-gui.bat
go-realtime-gui-new.bat
go-web.bat
install-easyGUI.bat
i everytime used go-web.bat
cuz i don't know what others do
-rvc
RVC Guides (How to Make AI Cover)
🇺🇸 English Documentation
Credits: AI HUB France
🇺🇸 English FAQ
Credits: AI HUB France
🍏 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
- 🇫🇷 French Documentation
- 🇮🇹 Italian Guides or
-it_g - 🇷🇺 Russian FAQ for RVC/Voice changer
- 🇧🇷 RVC Guides in PT-BR (
-rvc_br)
Whats the best RVC fork?
You'll have to try them
But Applio has a lot of features
While Mangio is more stable
I've been playing with Applio, but am running into lots of errors trying out the features
Bark-tts and UVR5 arent working. I'll have to play with Mangio
Try to use TROUBLESHOOT MODE or Make the Browser Privacy Setting to Custom and disable Cookies in Firefox
Alternatively use WaterFox, Librewolf or Mercury Browser (All Firefox Forks)
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for fairseq
Building wheel for antlr4-python3-runtime (pyproject.toml) ... done
Created wheel for antlr4-python3-runtime: filename=antlr4_python3_runtime-4.8-py3-none-any.whl size=141217 sha256=47fd144fe5849cb3d25956d9a121b3f5956e99b83a885364d3b9e313338a9eef
Stored in directory: c:\users\nilec\appdata\local\pip\cache\wheels\42\3c\ae\14db087e6018de74810afe32eb6ac890ef9c68ba19b00db97a
Building wheel for ffmpy (pyproject.toml) ... done
Created wheel for ffmpy: filename=ffmpy-0.3.1-py3-none-any.whl size=5601 sha256=e1bbbf7ce361acfabd13b628db6471d2acb206f2d87bddc339ca2c7c93fd6f52
Stored in directory: c:\users\nilec\appdata\local\pip\cache\wheels\1f\f1\8d\367922b023b526b7c2ced5db30932def7b18cf39d7ac6e8572
Building wheel for future (pyproject.toml) ... done
Created wheel for future: filename=future-0.18.3-py3-none-any.whl size=492054 sha256=9cea27b393717bc9aa709c51395aa73a35be83d60b098e0a815703344def0b34
Stored in directory: c:\users\nilec\appdata\local\pip\cache\wheels\bf\5d\6a\2e53874f7ec4e2bede522385439531fafec8fafe005b5c3d1b
Successfully built antlr4-python3-runtime ffmpy future
Failed to build fairseq
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects
whoops lol meant to nbot send it like that lmao but yeah any help?
I'll just warn you.
Do NOT run install-easyGUI.bat!
I'll repeat, do NOT run install-easyGUI.bat!
It may break Mangio...
okay
@serene hound has reached level 3.
thanks for the warning
Is it Applio-RVC you're trying to run?
Which area do I drag the file the vocals in Mangio to have the ai sing the vocals?
I try to upload the vocals, but it just gives me a sample audio of the model
@jaunty skiff has reached level 3.
You can't drag&Drop files in Mangio.
Right-Click the audio file and select "Copy as path".
Paste it in the 3rd field on the left.
Or place the audio file in the 'audios' folder and you can select from the list in the 4th field on the left (after refreshing voice list, index path and audios).
Can you send me a picture to where I should paste the vocals?
NOW HOSTING RVC HELP DESK IN MAX 15 https://discord.com/channels/1159260121998827560/1159291105423151104

@prime hemlock how to ai voice
yes
what do i do
@tidal summit has reached level 1.
if on huggingface
it keeps saying theres been an error
when trying to load a model in
Which HuggingFace?
RVC v2 Huggingface Version, by Clebersla
Huggingface Spaces
Should be a direct download link
Are you in the download model section?
yes
Did you post a direct download to the zip?
Do you have an account
You are probably missing 'buildtools' https://docs.applio.org/installation/requirements/
ight preciate it!
Do you mind if you send a picture for where I should paste the vocals?
i'm honestly so lost on what link to use to download rvc local and what guide to use because there are so many lol
Ok, Then what do I do?
Select the index file in the right next to it.
Select RVMPE or Mangio-Crepe for the pitch extraction algorithm.
Then you're good enough to press 'Convert'.
It's not finding the index file
Then you probably didn't put the index in the proper folder...
Where should I put the index file?
GordonRamsay.pth goes in the weights folder.
added_IVF*.index goes in logs/GordonRamsay
Where should I be able to find the "Logs/GordonRamsay" part?
You create the folder 'GordonRamsay' in the 'logs' folder.
Then you place the added_IVF*.index file in the 'GordonRamsay' folder.
Then you press the refresh button and it'll appear in the list.
basically, for yourmodel.pth, make a folder yourmodel in logs\ and put added_IVFXXXX_Flat_nprobe_1_yourmodel_v2.index inside it. I feel I've told you before...
When I click the arrow in the "Path to your added Index.file" It still doesn't pop up
Hit the refresh button and it should show in the list.
Got it!
Where should I put the vocals to a song?
Starting Applio with runtime for Nvidia support (you must have it installed)...
The system cannot find the path specified.
@wanton wraith has reached level 1.
doi need to add something to my path enviroment
See the screenshot. The highlighted text. There.
Or place the audio file in the 'audios' folder, hit refresh and select it from the list.
Ok, I think I got it. Thank you so much!
Launch 'go-applio-manager'.
Then '[3] Download NVDIA Runtime'.
One more thing, I try to request a model for a free request, but it won't let me post it
You need to select the 'Free Request' tag.
I tried doing that, It's not letting me post it
I clicked on "go-web" and this happened
you have another program using that port 7897
Nevermind I got it to work. How do I lower the pitch, because the voice sounds too high
Set a lower value for 'Transpose'.
Start with '-12'. That's an octave lower.
Ok, I'm a female and I recorded my voice to have him say stuff, so should I set it to -12?
Yes.
you guys are doing great work
when i launch the https_start file it will briefly open consol and then disappear how i get the rvc to actually launch. (dont tell me to reinstall ive done it 8 times)
@nocturne python has reached level 1.
Whenever I try to download model it says theres been an error
what kind of link
I'm trying to start applio with runtime on nvidia but it keeps saying "The system cannot find the path specified" and then I get sent back to the start runtime screen
oh wait i believe i h ave found the fix
@opal oxide has reached level 1.
I'm using applio version btw
@opal oxide make sure you've installed python 3.9 and added it to the system environment PATH, also the applio's directory path shouldn't have spaces
It just keep saying "could not build wheels for fairseq, which is required to intstall pyroproject.toml-based projects"
@snow hazel has reached level 3.
I have no idea to fix this and I went to many people and they don't know either.
use the runtime instead of trying to install dependencies by yourself
How do I do that. I'm kinda new to this.
open this and select Runtime (Nvidia support)
@proud elbow I've done that and this is what is gives me.
If i update my model on hugging face with another model (same name), will the wordband bot create new cover of previous covered song when i do the comand, or will it just send me the old song that it covered with the old model
most likely yes, that is if worldband downloads the model each time it makes a cover, otherwise it wont
i guess just test it
Kay ty
RVC local UVR shows this sometimes. why? it says there is no such file or directory... it literally has the file itself. why is this happening?
@visual pumice has reached level 2.
i need answers
because its abnormal how it only happens to some songs and doesn't happen to other songs. even with the same file format
so what the hell. what am i even suppost to do. its literally just rejecting the mp3 i give it
same
nvm fixed it the directory must have no spaces in the folders.
now i have another error... it says no-feature-todo
nvm i fixed it
how do i fix shit like this and get a better result
Does anyone know how I can make AI Voices raspy for scream/shout vocals?
fuck is going on with applio colab bruh
I think it's broken for real now
this ain't loading shit
I just wanna do some inference
i have trained a model to 150 epochs just as a first kind of test but when i try to use it in rvc gui it just tells me 'No eligible .pth files found in folder'
someone please help
let applio cook for a while, it does that 
Okay it's cooking
btw
am I gonna get disconnected after like
10 - 15 mintues
because it did that on the previous applio
probably 
😭
how long does processing usually take ?
its been like this for over 30 min for me
How long is your dataset?
like 20 minutes i thinkj all together
4 files
how many cpu processes are you using?
that's the issue
you shouldnt have changed that slider, can screw up preprocessing
so leave it at default?
or put it to 1
always set it to 2 regardless how many cpu cores. that also applies for feature extraction as well.
thank you!
@wanton wraith has reached level 2.
did you get disconnected?
No
any tips on how to speed up training?
Its fast enough but maybe someone knows how to make it even faster
im using an rtx 2060 6gb
batch size 3 performs optimally on a 6 GB vram card, or if you want it faster a 12 GB card can perform fastest on up to batch size 16
-rvc
RVC Guides (How to Make AI Cover)
🇺🇸 English Documentation
Credits: AI HUB France
🇺🇸 English FAQ
Credits: AI HUB France
🍏 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
- 🇫🇷 French Documentation
- 🇮🇹 Italian Guides
- 🇷🇺 Russian FAQ for RVC/Voice changer
- 🇪🇸 Español (
-rvces) - 🇧🇷 PT-BR (
-rvcbr)
sorry if this is a dumb question i dunno shit about using this but after i extract the model why is the zip folder empty
Did you click on Refresh?
yes
Try a restart of Applio or an automatic install of models using their links
(Resources > Download Model)
uh ok
non-ascii characters in the folder perhaps ?
True
But idk
I mean... Applio is still running
Wouldn't even start up
If it didn't support the characters
it might still run
I've seen most issues happen when python software started other processes
that's the issue with spaces : the initial script runs fine, but when another process is called, it invokes python, but it shits itself because it interprets spaces as multiple arguments rather than one path
might also fuck up with non-ascii stuff the same way
Put the the ModelName.pth files in logs/weights.
And the added_IVF*.index in logs/ModelName.
@wispy burrow has reached level 15.
yea that is in fact the fix
whats the difference betweem rmvpe_gpu and rmvpe and also whats the best model for uvr5 in applio?
help
Can't help if you don't give any information...
Do you still need help with that or the ModuleNotFound Error for 'torch'?
i have a question does anybody know a good model for removing piano from the background with both options ( with piano and without piano)?
for ultimate vocal remover 5
what a shame
Any good guides on data set creation? Trying to figure out how much audio to train with for a high quality model.
OOOOH NOOOOOO!
search on weights https://www.weights.gg/
Hi everyone! There's been a recent update I guess with RVC Disconnected and I'm not sure where I'm going wrong since I've been doing fine so far until this update:
|
fatal: destination path 'Mangio-RVC-Tweaks' already exists and is not an empty directory.
OSError Traceback (most recent call last)
<ipython-input-29-bfb8a4688ddf> in <cell line: 13>()
11 get_ipython().system('git clone -b pr-optimization --single-branch https://github.com/alexlnkp/Mangio-RVC-Tweaks.git')
12 #Rename to keep backwards compatibility with old variants of Disconnected
---> 13 os.rename("/content/Mangio-RVC-Tweaks", "/content/Mangio-RVC-Fork")
14 get_ipython().system('git clone https://github.com/maxrmorrison/torchcrepe.git')
15 get_ipython().system('mv torchcrepe/torchcrepe Mangio-RVC-Fork/')
OSError: [Errno 39] Directory not empty: '/content/Mangio-RVC-Tweaks' -> '/content/Mangio-RVC-Fork'
10 - 50 minutes of audio (without silence or any unwanted noises).
The more the better.
JK i just can't read
ayo, what the heck is this and why? somebody? haha
i just restarted the machine on paperspace, ran everything up again and i saw this
Anyone get Tensorboard to work with the runtime version of Applio?
@hardy idol has reached level 1.
How i make a cover with the interface of this colab?
https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=Fz3XSI8GrXra
Im not a Premium user
Yes
You can't
Use an alternative listed in #1159949278270193734
Use the Scalars tab
Oh lol
does more steps equal better result? Or more epochs and less steps is better?
can i get some help when i use the rvc local client i get this error hod: crepe file_index: models\kurt cobain\added_IVF884_Flat_nprobe_1_v2.index file_big_npy: index_rate: 0.4 output_file: C:\Users\codyn\Downloads\Foo Fighters - Best of You\vocals_RVC_1.wav
Initiating prediction with a crepe_hop_length of: 128
Using the following f0 method: crepe
Traceback (most recent call last):
File "D:\RVC-GUI-pkg\rvcgui.py", line 129, in vc_single
audio_opt = vc.pipeline(
File "D:\RVC-GUI-pkg\vc_infer_pipeline.py", line 340, in pipeline
self.vc(
File "D:\RVC-GUI-pkg\vc_infer_pipeline.py", line 243, in vc
(net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0])
RuntimeError: CUDA error: the launch timed out and was terminated
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.
i have installed applio, i have 2 EXE's. do i run ffmpeg, or ffprobe?
i have ran install_applio.bat
install runtimes
more steps do not equal better results its just the same
install what?
so i just made a .pth file, but it's not valid, or at least that's what it says. what did i do wrong?
runtimes
im so confused 😭
can i get some help other here?
@finite cobalt has reached level 4.
after installing applio, it should say done and the folder will be there
then you open go applio manager.bat and fro mthere install the runtime that fits your GPU
you wait until its over
then you do go-applio and start the server
and go ahead with your training
any questions? @formal chasm
wdym you "made" a .pth file?
how to use ckpt processing for combining two voices into one ?
easy gui that is certainly a big no-no
and got a .pth file
how tho
where do i get the link from
hi im having an issue, everytime i post the link into the gui it asks for a mega link
#@markdown Link the URL path to the model (Mega, Drive, etc.) and start the cod
what gui?
be more specific
Easy GUI (for RVC v2, with crepe) (with improved downloader)
another easy gui user sigh
im very new to this found it through a youtueb vid
don't use easy gui, use mangio or applio
hey dont hate everyone starts somewhere
am not hating, just saying
would you mind sending the link
@lucid spire has reached level 1.
alright.
thank you
as in website or local training?
i quit
that's what all gamblers say before winning big
do you have any resource material on how to attack this lmao
attack? as in clash of clans attack?
nah more like brawlstars
but nah like wtf i do haha the easy one was v simple but didnt work
im tryna make a spongebob banger
rightttt
but like i have no experience or understanding of coding, not your job to teach me lol was just looking for a step by step
I don't recall mentioning the word "coding" anywhere?
that is a step by step though lol. Download and open
so i openned the link you sent
and now im just lookin at stuff
do i hit go to model
pretend im 5, where do i click
thank you for helping btw
okay i downloaded it onto my computer so now i have the folder
code --> download as zip
did you download the zip?
yes maam
unpack it
done
what python version do you have installed?
ha none i will do that now which one should i instal
3.10
can you show a more detailed error?
screenshot or copy paste
just got python
@lucid spire has reached level 2.
give me a link
im so lost even opening this
website
it doesnt even have to have all the functionality kitsai has
@magic hare has reached level 1.
just a simple voice model training website or smth will do
one that doesnt get all technical yknow\
I run Applio installer and this happened! what did I do? It was working fine before what happened?
You don't have permission to access this resource.
Apache/2.4.38 (Debian) Server at frippery.org Port 443
At line:1 char:1
+ Invoke-WebRequest -Uri https://frippery.org/files/busybox/busybox.exe ...
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo : InvalidOperation: (System.Net.HttpWebRequest:HttpWebRequest) [Invoke-WebRequest], WebExc
eption
+ FullyQualifiedErrorId : WebCmdletWebResponseException,Microsoft.PowerShell.Commands.InvokeWebRequestCommand
'busybox.exe' is not recognized as an internal or external command,
operable program or batch file.
Extracting the runtime_dml.zip file...
Exception calling "ExtractToDirectory" with "2" argument(s): "Could not find file
'C:\Users\brand\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\runtime_dml.zip'."
At line:1 char:63
+ ... ileSystem ; [System.IO.Compression.ZipFile]::ExtractToDirectory('runt ...
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo : NotSpecified: (:) [], MethodInvocationException
+ FullyQualifiedErrorId : FileNotFoundException
Could Not Find C:\Users\brand\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\runtime_dml.zip
The system cannot find the path specified.
Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases.
Applio has been successfully downloaded, run the file go-applio.bat to run the web interface!
Batch process completed with exit code 0
Hello, anyone please help?
res is too high in game
For the Fairseq issue, you are probably missing the 'Buildtools'.
https://docs.applio.org/installation/requirements/
Download 'buildtools' and when installing check those first two on the left.
You can look at the image to see which ones.
For the ModuleNotFound error for 'torch'. (I assume you have an NVIDIA GPU) pip uninstall torch torchvision torchaudio pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117`
That should fix it. (It did for me)
why is my tensorboard only capturing data every 200 steps on the latest release of rvc?
They have it setup that way
I don't recall which file needs to be edited to change it though
oh thanks, is there reason for that?
I really don't know, maybe it made training much slower on some systems ? Seems silly ¯_(ツ)_/¯
I feel like i'm missing something. But I don't know what
I'm stuck
@brittle wing has reached level 2.
For some reason the file isn't available to you
Not sure how to fix it
The installation script seems to do a lot of wacky shit for god-knows-what reason
That is really odd. I have no idea why does this all of a sudden. It was fine when I last installed it
Does the following website download a file just fine ?
https://frippery.org/files/busybox/busybox.exe
(that's the file it's trying to download, and on which it fails)
Oh okay. I thought that was an application to generate the error log
Lol
Let me try to download it, maybe it will work
I don't know which folder it'll need to be put in
but it's even weirder if it downloads from browser but not from the script
Oof
I have visual studio build
also sorry for late reply
But did you select those 2 things when you installed it?
What two things?
Okay time for me to screech about applio again
It didn't tell me to install anything
WHY THE FUCK IS IT ALWAYS STUPID
This is the first time I'm doing this. I have no idea what to do.
I swear the more I see applio the more i see dumb shit they get away with
So I don't need any of this?
I think your error has a message along the lines of "please install the c++ build tools from Microsoft"
I have that though.
And some specific components of it should be installed as well, for it to work. There's exactly two components needed
can you screenshot the tickboxes for the c++ build tools ?
My screenshot with the correct choices are gone with the old server
Hi! How much epochs are recommended for a 15 minute dataset??
"Normally" you'd need Git and the Applio would do the rest...
But it doesn't for some, if not most of the people...
I don't see tickboxes?
I don't see that
See this image ? (don't follow the same settings)
That seems to be the wrong software
That what I got when I pressed the link
Mind sending the link ? I don't recall it linking to that
The link that was in https://docs.applio.org/installation/requirements/ ?
I pressed buildtools
Don't fucking tell me the applio docs point to the wrong thing
Is it the same for this one? https://us.aihubfrance.fr/vocal-cloning-guide/applio
Yep
pfff...
I don't have a windows machine to test atm
But it could be the right file, not sure what's up with it
Just gonna put this here before I go to bed : The only required components from the build tools are :
The C++ build tools, and within the C++ tools, also select "Windows 11 SDK" and "MSVC"
that's it
It's some more utilities that are needed for some python stuff to build correctly
The Installer I had now shows that too...
It says to update it...
@snow hazel has reached level 4.
"Desktop development with C++"
&
.NET desltop build tools".
".NET desktop build tools".
Reeeee .net isn't needed
Okay...just "Desktop development with C++" then?
All of this good?
I mean you selected all of them, it'll work, but it's going to take a while to download
and use a whole lot of storage space
You can untick all of them but those two :
- MSVC v143 - VS [...]
- Windows 11 SDK
I didn't select anything. It's defalt
Uh it’s been stuck for a while now?
Wait what it just changed to 1 of 18
Alright then….
How much should I set this if my gpu has 6gb of vram?
And which one do I do?
Keep it default
Also I might of been mistaken. I think it was on the right thing
"Get Applio. It's better!"
"Everything is better with Applio!"
But an installer that actually works and a proper guide with correct links so it doesn't download/install unnecessary things? Too much to ask?
The link for Applio in #1159949278270193734 links to Applio Installer (V2)...
Even though there's Applio Installer (V4.01)?
Don't get me started on applio lol. All that PR about being user friendly...
Yet I've never seen as many issues
any of the two should work
sorry for the delayed response lol
Do I need to sign in to visual studio?
No, shouldn't be needed
Ok
IT FINALLY WORKS!!! THANK YOU GDR AND RANDOM BLOKE!
how much epoch should i use for a 7 mins 16 seconds dataset
Okay I downloaded it, and I have no idea what to do now. It still gives me a problem
I imagine you saying No the luigi meme way
Whoever dm'd me for help, accidentally hit ignore instead of accept 
Nani? O.O
Damn... no solution to my problem?
Oh well
I'm trying to convert on huggyface it appears the error of the audio
Halp! I tried to train my own voice model but there's this weird glitchy noise whenever I infer! https://voca.ro/1lLfMCJuu4tk What's happening?! How do I fix it?!
(note: the text at the end is p12_i0.5_fr3_rms0.25_pro0.33_rmvpe, song title and voice name i'd rather not say yet)
It's not a problem with RVC or the GUI I'm using, because other models work just fine.
sound like it wasnt trained properly. Can you tell the dataset size, how long it took to train, batch size and epoch and steps?
I have a voice model all working, even made some song covers with it but I can't seem to find a TTS or maybe I'm just dumb, could anyone help direct me to one?
The dataset was about 4 mins long, 1000 epoches and took about maybe 2 hours to train. idk the amount of steps
@ornate ice has reached level 1.
overtrained?
No! Even when I trained it at 10 steps, the robotic noises were present as usual!
What’s the difference between training 300 epochs to 5000 epochs or 10000 epochs?
Higher the better/accurate singing vocals?
Idk if this is true for all model, but if you overtrain it, then the voice gonna start to distorted, especially at high note, so it best to stop at the lowest loss
Does the amount of Datasets, such as 1 min, 10 mins or 20 mins of Audio Datasets affect Epochs?
Assuming the higher the mins of Datasets meaning higher of Epochs do better?
Yea the bigger the dataset, the higher the epoch is
how long does it take to infer a 5 minute song w pm
currently its at 400
@solar dawn has reached level 1.
ok my pc deadlocked
can someone help with this how tf does my download fail in LOCALHOST
@brittle wing has reached level 1.
I wonder, how many Epochs should I do for a 6:44 minutes of Datasets?
Im havent really done this training stuff much to know which is the optimal epoch, best thing to do now is try to test with a big epoch, then look for the g loss graph on tensorboard to see the optimal epoch
Like if the graph went down, it mean it still good to train and can still be improved, if it start to go into a flat line then that mean it overtrained, the optimal stop would be the lowest point of the graph
Is there a chance that the laptop using 4GB RAM and AMD Ryzen 3 3200U can use Mangio RVC Fork locally?
how do i make this work?
No, that's too little unfortunately
oooof
amd support already isnt great
and with just 4gb of ram and on an integrated gpu?
sorry fam, not happening
try pip install sensorboard
pip install tensorflow
tensorboard*
just use low
better for performance anyway
Plz help
just to confirm, https://i.nekogirl.net/0d168720b963 this means RVC V2 right
Yes it's RVC v2
got it
can i have the colab link, if that even exists by now.. if not, the way you can do it on a web browser
-colab
Google Colabs
RVC
- AI-Cover-Gen-No-UI (inference only)
- RVC v2 Disconnected (training only)
🌐 Translated colabs
- PT-BR:
-colabs_br
Audio Separation/Isolation
thanks
@merry spindle has reached level 1.
experiment name: the name of the voice you're making, or whatever. it's an experiment i think
model_architecture: the model you are using to train the voice. maybe set this to v2.
target_sample_rate: the sample rate of the voice. maybe set this to 48k.
speaker_id: RVC only allows one speaker at a time. best leave this to 0.
pitch_extraction_algorithm: i will elaborate further on this one.
crepe_hop_length: i have no idea. leave this to 64.
pitch_guidance: LEAVE THIS TO "yes". i think.
which RVC client is the best for API usage?
what is cpu threads
do i just put the audio file in that folder..?
please answer fast
cpu threads is the amount of aeuuughhhh your rvc uses. i have no idea. best set this to 2.
create a zip file using that format
then create a folder called "rvcDisconnected" in your Google Drive. dump the zip there.
then tell the notebook the name. it will look in /rvcDisconnected.
i already i have an rvc disconnected folder in my google drive
@merry spindle has reached level 2.
@winged comet has reached level 2.
am i allowed to use anything for the filenames
what do you mean?
just to be safe, go like "file1.wav, file2.wav, file3.wav" and so on.
convert it to a Wav just to be safe.
okayy
for example, if your character's name is danhamnthonybamflamthonanieltriceratruckwiddlesworth-rodriguez, make a folder called "danhamnthonybamflamthonanieltriceratruckwiddlesworth-rodriguez", then dump your wav files in there, and then create a zip file with the folder in it, name it whatever you want, and put it in rvcDisconnected.
@ornate ice has reached level 2.
thanks
am i allowed to take as much time as i want doing this?
like is colab gonna break
yes you may.
HOWEVER
if you train for too long...
then colab's back...
will CRACK
is only having one wav okay??
yes
how do i make zips again
one. right click on a file in File Explorer
please answer the question
what question
this
very yes
NO DON'T EXTRACT IT
you can name it anything
but make sure to input the right name here
@merry spindle
okay
i didnt
i imported the dataset
what do i do now
@ornate ice
so you loaded the dataset?
do i just click preproccesing
wdym
@merry spindle has reached level 3.
ok do all these,
including save preprocessed dataset files
but SKIP THIS, you don't need it
i started training!!!!
and skip this too
this new colab is VERY confusing to me
only do the (for resuming) sections if something unexpected happens and colab's back cracks
you do these to continue the training
the old one was easier
this is under the thing showing me the epochs that are training
why does appilio need internet to open locally lmfaop
probably they need to download something from the web
@ornate ice has reached level 3.
good to hear
oh btw, how fast does it train 1 epoch
WARNING: watch for new files that appear on the drive. it regularly spits out files, it could clog up your drive
huh? what's happening?
its going very fast
i think something is wrong
did it go from very slow to suddenly very fast, or did it slowly speed up
everytime
i open it
maybe wait for the internet, it's the only thing you can do
sometimes its fast, sometimes its slpw