Online or locally?
#✨│ai-help
brittle wing
this? is it?
low shard
this? is it?
this is the colab https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sharing
this is the guide of the ilaria rvc colab
gritty tundra
Online or locally?
I want to do it online I'm just an open person
brittle wing
That's all. There's no step
brittle wing
this is the colab https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMj...
Thank you so much. The solution is complete!
gritty tundra
check out this https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnecte...
I had my own vocal model in mind so that I could have my own
low shard
That's all. There's no step
i meant in the doc guide
low shard
Thank you so much. The solution is complete!
yw
gritty tundra
check out this https://aihubdocs.github.io/en/rvc/cloud/training/rvc-disconnecte...
good
low shard
do not use webUI if you don't have the paid plan
it was the old rvc easy gui improved downloaded, so it doesnt work in general
low shard
I had my own vocal model in mind so that I could have my own
yea that one is for training
gritty tundra
yea that one is for training
How to make your own RVS model
proper shale
How to make your own RVS model
https://aihubdocs.github.io/en
Has everything you need to know
gritty tundra
https://aihubdocs.github.io/en
Has everything you need to know
but it's not in Russian
And here you can make your model local
low shard
but it's not in Russian
sorry but i dont speak russian, i don't know how to translate you the guide
low shard
And here you can make your model local
it depends if your pc is good, if it isnt udo it with the rvcdisconnected colab,online
azure marshBOT
RVC Guides (How to Make AI Cover)
gritty tundra
it depends if your pc is good, if it isnt udo it with the rvcdisconnected colab,...
I have a GTX 1050 video card
low shard
I have a GTX 1050 video card
its better if you do it online with the rvcdisconnected colab then
sorry but idk how to translate the guide in russian tho
marsh fractal
python trainset_preprocess_pipeline_print.py "/content/dataset/voiceproject" 32000 2 "/content/Mangio-RVC-Fork/logs/voiceproject" 1
python3: can't open file '/content/Mangio-RVC-Fork/trainset_preprocess_pipeline_print.py': [Errno 2] No such file or directory
python extract_f0_print.py "/content/Mangio-RVC-Fork/logs/voiceproject" 2 rmvpe 64
python3: can't open file '/content/Mangio-RVC-Fork/extract_f0_print.py': [Errno 2] No such file or directory
python extract_feature_print.py "device" 1 0 0 "/content/Mangio-RVC-Fork/logs/voiceproject" v2
python3: can't open file '/content/Mangio-RVC-Fork/extract_feature_print.py': [Errno 2] No such file or directory
Hello! Sorry I dont really know what Im doing. Can anybody help me out with this? Im following the step by step process in the website
I cant start Index Training. It says there are no features for the model yet
low shard
I cant start Index Training. It says there are no features for the model yet
it says no feature todo right?
its prolly a problem with the dataset
check that is in
.wav format
and also how you made it?
marsh fractal
its in .flac
low shard
.wav is the most compatible format, its better you extract now the dataset in audacity in wav or just convert it to wav online
marsh fractal
alright, thanks. Ill try that
low shard
yw
marsh fractal
Ive changed it but it doesnt seem to work still
rare gobletBOT
Ayo? @marsh fractal level 1 !!! 
low shard
Oh, if you got any errors im here
split latch
cleaning up Charlies dataset
let me know what you think about it
split latch
thanks!!
I might as well start training
hexed cave
still crystal
I have a GTX 1050 video card
i have gtx 1050 its not good for training
gritty tundra
i have gtx 1050 its not good for training
Why is this so?
rare gobletBOT
Ayo? @gritty tundra level 8 !!!
still crystal
Why is this so?
i dont know its just not good
gritty tundra
i dont know its just not good
UwU
iron surge
rvcs dont work for me lol
rare gobletBOT
Ayo? @iron surge level 1 !!!
rotund prawn
its applio
half cove
is fcpe better than rmvpe
Fcpe is a bit faster but rmvpe is better quality and fast
Just use it
low shard
rvcs dont work for me lol
!howtoask
iron surge
damn
low shard
Be specific when asking for help
covert dome
yo guys, anyone could help me to put my ai voice on discord ?
because it doesn't detect my microphone
i really need help to understand, cause it doesn't work
severe sand
Can I just add files to the logs, to have them included in training?
gusty kestrelBOT
@covert dome, I have found 1 results that match your search!
Requested by lamasse.
half cove
@covert dome #🔍│find-models
covert dome
<@847774560668745738> <#1163592055830880266>
my bad
rare gobletBOT
Ayo? @covert dome level 1 !!!
brittle wing
What' the difference between RVMPE & rvmpe_gpu
sour mason
for training, I found the progress in logs folder. There are a bunch of D and G pth files. which one do I choose? also do I use added or trained index file?
rare gobletBOT
Ayo? @sour mason level 4 !!!
lavish lintel
Hello how long does it take to train i have ryzen 7 3700x rtx 3070Ti(8gb) and my dataset is a little under 4 minutes
low shard
for training, I found the progress in logs folder. There are a bunch of D and G ...
did you finish training everything?
when its finished, you need only the added index file and the (modelname).pth, you just need to zip them
sour mason
did you finish training everything?
not yet but I set the goal to be 1000 cuz the guide tells you to
so the plan was to go on until OT point and stop
low shard
yea, wait till that and youd need the files i told you
sour mason
by till that do you mean the OT point
low shard
yea
sour mason
yesterday when I stopped training, I couldnt find any modelname.pth
is it saved somewhere else?
severe sand
More data = better? or is there a point where more makes it worse? (ignoring that it takes longer to train)
low shard
yesterday when I stopped training, I couldnt find any modelname.pth
it should be in the weights folder
sour mason
it should be in the weights folder
where can I find this weights folder?
low shard
More data = better? or is there a point where more makes it worse? (ignoring tha...
yea there is that point, there is undertraining and overtraining, you can check up more about this at https://aihubdocs.github.io/en/rvc-resources/epochs-overtraining--tensorboard/
sour mason
it's not in my logs/model folder
severe sand
where can I find this weights folder?
RVC1006Nvidia\assets\weights
sour mason
rn it's not there so Im assuming when I stop training, it will pop up there?
severe sand
yea there is that point, there is undertraining and overtraining, you can check ...
That site (under datasets) recommends to not include strong S SH and P sounds. Doesnt that remove the ability of the model to make those sounds?
Or are they only refering to very extreme sounds like when you almost swallow the mic?
low shard
it's not in my logs/model folder
its not in the model folder, should be in the weights one
low shard
rn it's not there so Im assuming when I stop training, it will pop up there?
do u see smt like (modelname)_e(number)_s(number).pth ?
this is an example on colab cus i dont do it locally lol
low shard
That site (under datasets) recommends to not include strong S SH and P sounds. D...
its so that the model doesnt do weird sounds, it needs to have a clean dataset of the voice
light meadow
small question : what setting in ColabGen No UI (google colab for making ai cover) so that it don't merge the reverb and the ai vocal ?
i mean i want the 100% ai cover together with instrument
severe sand
its so that the model doesnt do weird sounds, it needs to have a clean dataset o...
Well I'm currently trying to add to the ability of my model, since it cant do any noises like mmh sounds, grunts or whatever. Someone said that I should include some in the dataset but now reading that I'm unsure if that will work
Guess I will have to try
low shard
Well I'm currently trying to add to the ability of my model, since it cant do an...
you really dont have to add those grunts or mmh sounds tbh, btw who told you to add those
low shard
small question : what setting in ColabGen No UI (google colab for making ai cove...
you mean ai covergen no ui?
light meadow
you mean ai covergen no ui?
ye it is
molten pecan
@low shard @half cove @viscid moss big update is coming...
severe sand
you really dont have to add those grunts or mmh sounds tbh, btw who told you to ...
Well the model is completely incapable of doing them though? If I just go "mmmh" (like thinking) it will do some random vocals and often add a random S noise to it.
just doing whatever else
How would you go about solving that?
low shard
Well the model is completely incapable of doing them though? If I just go "mmmh"...
wait you mean you already trained it and it goes like that?
so you are re training it to fix it?
severe sand
wait you mean you already trained it and it goes like that?
Its already trained and I would change the dataset and try again yes. I haven't done anything yet though, still trying to figure out if thats the right way to go about it.
Its a super clean dataset and the model itself sounds good, just these noises are something it cant do at all
light meadow
so you want it to merge or not?
erm i don't want it merge the AI vocal with reverb, i just want AI cover merge with instruments
also don't even need to split main vocal and back vocal i think
low shard
Its already trained and I would change the dataset and try again yes. I haven't ...
ohhh, got it now, weird never got that type of error, maybe you could try adding them, but dont know much as usually it works all fine for me
severe sand
ohhh, got it now, weird never got that type of error, maybe you could try adding...
So when you do something like "umm" "mmh" it all translates well? I mean its not like my dataset doesnt have any m's in it, just no exact prolonged sounds like that.
rare gobletBOT
Ayo? @severe sand level 5 !!!
low shard
erm i don't want it merge the AI vocal with reverb, i just want AI cover merge w...
sorry but i dont use the aicovergen no ui colab, maybe @viscid moss could help you
low shard
So when you do something like "umm" "mmh" it all translates well? I mean its not...
usually i dont have those types of problems, its kinda weird, you could also make a post on #1192011222023950368 about this cus its the first time i hear smt like this
viscid moss
erm i don't want it merge the AI vocal with reverb, i just want AI cover merge w...
download the file that ends in rmvpe (but not the one that ends in mixed) and download the instrumental. Then you join them again in a program like audacity
light meadow
ty :3
low shard
Yea
Thst wsy you will only get the modelname.pth file at the end of training, but you cant do nothing if the model overtrained before the training finished, so see how it goes
severe sand
usually i dont have those types of problems, its kinda weird, you could also mak...
Made a post and added a sample
https://discord.com/channels/1159260121998827560/1209181426663301161
sour mason
makes a lot more sense
proper shale
delete stored_setting.json inside the MMVCServerSIO folder, restart
else it might be the av/firewall messing up, if you have any of thesw
brittle wing
delete stored_setting.json inside the MMVCServerSIO folder, restart
alr thx works now
stray sage
How do I know if my gpu is used?
it stays at 2% utilization when cpu reaches 90 to 100.. i don't think it's using my gpu..
proper shale
yw :)
iron surge
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "voice_changer\RVC\pipeline\PipelineGenerator.py", line 30, in createPipeline
File "voice_changer\RVC\embedder\EmbedderManager.py", line 26, in getEmbedder
File "voice_changer\RVC\embedder\EmbedderManager.py", line 51, in loadEmbedder
File "voice_changer\RVC\embedder\FairseqHubert.py", line 11, in loadModel
File "fairseq\checkpoint_utils.py", line 425, in load_model_ensemble_and_task
state = load_checkpoint_to_cpu(filename, arg_overrides)
File "fairseq\checkpoint_utils.py", line 315, in load_checkpoint_to_cpu
state = torch.load(f, map_location=torch.device("cpu"))
File "torch\serialization.py", line 797, in load
with openzipfile_reader(opened_file) as opened_zipfile:
File "torch\serialization.py", line 283, in __init
super().init(torch._C.PyTorchFileReader(name_or_buffer))
RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory
[Voice Changer] pipeline create failed. check your model is valid.
[Voice Changer][RVC]: update_settings serverReadChunkSize:384
[Voice Changer][RVC]: update_settings serverReadChunkSize:384
[Voice Changer][RVC]: update_settings extraConvertSize:131072
[Voice Changer][RVC]: update_settings extraConvertSize:131072
[Voice Changer][RVC]: update_settings modelSlotIndex:1708363989001
[Voice Changer][RVC]: update_settings modelSlotIndex:1708363990001
rvc waiting for pipeline
sacred lichen
can someone help me find a guide for advance setting RVC realtime
some setting are very hard to understand
brittle wing
same problem here
boreal axle
reinstall
rare gobletBOT
Ayo? @boreal axle level 2 !!!
boreal axle
fixed it for me
sacred lichen
<#1159290161683767298>
no not okada, RVC
molten pecan
The okada channel is for realtime inference too
brittle wing
On the trainer, do I look at the lowest point of the graph & look at "Smoothed"???
Then I know which epoches am choosing lol
fleet inlet
my okoda rvc wont use my gpu
azure marshBOT
RVC Guides (How to Make AI Cover)
molten pecan
On the trainer, do I look at the lowest point of the graph & look at "Smoothed"?...
Keep the smoothing at 0.987
Hover your mouse over the lowest point
A black rectangle will appear showing the traits of that point
See the Step number. Use the checkpoint that was closest to before the overtraining point
The checkpoints will be organized with this format: ModelName_Epoch_Step.pth
Example: julia_e60_step120.pth
You got it now?
@iron surge
molten pecan
my okoda rvc wont use my gpu
brittle wing
You got it now?
Understood 🙂
iron surge
<@598510216564310045>
brittle wing
fixed it for me
thx bro, fixed
fleet stirrup
-colab
wispy lion
So i clicked "train feature index" and nothing happens~, not being able to get index file even after training it
sinful iris
rugged meadow
why does every model i make sounds the same and with the same voice
unkempt valve
I've already tried a lot of things, can anyone help?
carmine oar
anyone able to help me set up Python for rvc web-go bat to work?
valid spruce
Hey guys! can you help me? I wanted to know how I can retrain my model
mental fox
hey guys, i'm looking for a billie eilish model that works, for some reason I keep getting this error: ModuleNotFoundError Traceback (most recent call last)
<ipython-input-5-b114748cdea6> 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'
rare gobletBOT
Ayo? @mental fox level 1 !!!
odd shale
Hey guys! can you help me? I wanted to know how I can retrain my model
Do you still have your training files like the folder of your model that was made on the logs folder?
You remember your training settings?
proper shale
hey guys, i'm looking for a billie eilish model that works, for some reason I ke...
you're probably using an outdated colab...
@half cove pls give ila rvc fixed colab 😭
half cove
<@936673139419664414> pls give ila rvc fixed colab 😭
isnt the normal fixed now?
proper shale
eh idk
half cove
eh idk
it uses the numpy fix instead of changing the code tho 
so longer install than mine
proper shale
TUTEL
mental fox
you're probably using an outdated colab...
Thanks very much. Do I just install and then copy and paste voice-model .zip into Tunnel: Gradio ?
odd shale
Thanks very much. Do I just install and then copy and paste voice-model .zip int...
No
Do it on the Gradio GUI
On the Download Voice Models tab
proper shale
^
mental fox
is that in the colab or in this discord server?
The new colab I just got only has install and start options, so I'm pretty confused
odd shale
is that in the colab or in this discord server?
On the Colab
wispy lion
So i clicked "train feature index" and nothing happens~, not being able to get i...
someone help? pls?
wispy lion
i think i have
proper shale
also logs folder
wispy lion
logs from the voice i'm trying to do , there is no "added" index file
proper shale
hm, it did train the model
wispy lion
whenever i press this it doesn't do it
carmine oar
is anyone able to help me? i dont know how to set up python proporly for rvc webgo.bat to work. it will give me a page that looks like
C:\Users\User\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
'runtime\python.exe' is not recognized as an internal or external command,
operable program or batch file.
C:\Users\User\Downloads\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0>pause
Press any key to continue . . .
someone people have told me its a python issue. im using this on a windows laptop.
odd shale
logs from the voice i'm trying to do , there is no "added" index file
You didn't train the index
wispy lion
nothing happens
wispy lion
You didn't train the index
that's the problem
when i press train index nothing happens
ripe grove
that's the problem
what does the terminal window say?
odd shale
when i press train index nothing happens
You can retrain your model on Mangio Fork
wispy lion
You can retrain your model on Mangio Fork
that was very buggy for me the voice sounded the same everytime
ripe grove
that was very buggy for me the voice sounded the same everytime
what do you mean It sounded the same?
wispy lion
every time i trained something it always sounded the same
the results were the same
and now that i'm not getting this "index" file i'm having the same problem
it was working just fine yesterday but today its not
i press this and nothing happens , in the cmd doesn't even say any error
ripe grove
it was working just fine yesterday but today its not
are you saying that when you used your model it didn't sound like the model you trained?
wispy lion
are you saying that when you used your model it didn't sound like the model you ...
yup
i really wanna fix this
carmine oar
whats the propor rvc down load for windos 11
molten pecan
@carmine oar read my bio
carmine oar
oh im sorry
@molten pecan i use windows 11, what does it mean when i have a screen that shows press any key to continue?
molten pecan
<@791013455251374151> i use windows 11, what does it mean when i have a screen t...
iunno
I don't use RVC locally
Ask @half cove
half cove
<@791013455251374151> i use windows 11, what does it mean when i have a screen t...
show image
im guessing it says somthing about it being paused
sour mason
is this ok? epoch is already at 680
sour mason
ok it's just that I expected it to OT around 500
carmine oar
@half cove do you know what i can do about this
rare gobletBOT
Ayo? @carmine oar level 2 !!!
half cove
<@936673139419664414> do you know what i can do about this
what did you use to unzip it?
carmine oar
winrar
carmine oar
is 7zip somehting i need to download seperate?
carmine oar
when i download Mangio-RVC-v23.7.0_INFER_TRAIN.7z
from the site it keeps downloading through winrar and not 7z
do we do screensharing help in this server?@half cove
half cove
when i download Mangio-RVC-v23.7.0_INFER_TRAIN.7z
from the site it keeps downlo...
no you use 7zip to unzip it
carmine oar
ik
ivory arch
Anyone get this error? [Errno 2] No such file or directory: './training//train.txt'
Using AI-voice-cloning when trying to train
carmine oar
it would help if i could screenshare since youd be able to see what im talking about
main lichen
Everything is running smoothly
But i cant seem to hear my voice
Ofc i dont use this voice its just an example
But when i press start nothing plays
Anyone??
@molten pecan
Uhuh
Yeah
I already have it
It worked but randomly it just stopped working
hushed gust
It worked but randomly it just stopped working
check these
i cant send pic
check the in and out
main lichen
Yeah i selected it in output
molten pecan
<@791013455251374151>
idk how to use Okada. ask in #🔍│help-w-okada
main lichen
alr did
hushed gust
idk how to use Okada. ask in <#1159290161683767298>
what's the difference between them tho
molten pecan
what's the difference between them tho
this one is for datasets, setting up RVC, training or inference through audio files
the other is for real-time conversion
main lichen
Yeah i gtg in 5 mins i really need help rn
hushed gust
the other is for real-time conversion
how can i get the other one <:
rare gobletBOT
Ayo? @hushed gust level 1 !!!
molten pecan
how can i get the other one <:
https://aihubdocs.github.io/en/ for inferencing through audio files? here's everything you need to know
use Mainline
hushed gust
https://aihubdocs.github.io/en/ for inferencing through audio files? here's ever...
noo i meant the real time conversion
hushed gust
i have the same as this one
molten pecan
noo i meant the real time conversion
you wanna know how to download okada?
i don't use it so idk
ask in #🔍│help-w-okada
carmine oar
any other admin able to help me?
brittle wing
I've not overtrained & I know how to make voice models, but if I use it for my voice changer, it can be obvious that it's robotic. I have given the trainer some time & I pick the lowest best results before the overtrain!
azure marshBOT
RVC Guides (How to Make AI Cover)
brazen fossil
Can someone help me, whenever I talk into my mic I just get this weird audio:
https://cdn.discordapp.com/attachments/1163976901942059138/1209286895897415710/2024-02-19_16-53-17.mkv?ex=65e65f2f&is=65d3ea2f&hm=2acb3f3ac5c58d65ad4a09b03033a12ec93ce0027543bfe3c15018b7e12392b2&
https://cdn.discordapp.com/attachments/1163976901942059138/1209289344167186494/help.png?ex=65e66177&is=65d3ec77&hm=014c86a187a0043b33325db2d14a24c53d6becca4a89bc81e029a35b272bd0d3&
(I had to leave download links as I could not attach files to this discord server for some reason)
outer sedge
# RVC Guides (How to Make AI Cover)
i aint reading allat
molten pecan
guys, how to option this
brazen fossil
<#1159290161683767298>
that takes me back to my other message, sorry I sent two
molten pecan
check the AICoverGen page
brazen fossil
check the AICoverGen page
Where can I find that?
light meadow
https://aihubdocs.github.io/en/
i read it but have nothing ;-;
brazen fossil
can you send a scs?
of what...?
molten pecan
i read it but have nothing ;-;
can ou send a scs?
brazen fossil
of your Colab space
What's colab space
light meadow
can ou send a scs?
ummm screenshot of what '-'
molten pecan
ummm screenshot of what '-'
of your Colab space
brazen fossil
check the AICoverGen page
What's the URL for that...?
molten pecan
Where can I find that?
brazen fossil
https://aihubdocs.github.io/en/
I can't find anything on w-okada
light meadow
I do see Audio Mixing Options
yes, i mean how to use it
i mean how to options the numbers
'-'
fast gate
guys istg
rare gobletBOT
Ayo? @fast gate level 2 !!!
fast gate
i try to use the ilaria rvc colab
and it just gives me error everytime
what could i be doing wrong
dusty rampart
i try to use the ilaria rvc colab
check if you skipped any cells
and send a screenshot of the error
otherwise we can't really help you
fast gate
ok first
dusty rampart
fast gate
ok now
i put the pth in the weights folder
and the model doesn't load as much as i refresh
hallow ferry
Someone could tell me where the problem is?
molten pecan
Someone could tell me where the problem is?
molten pecan
what could i be doing wrong
Did you read the docs guide?
https://aihubdocs.github.io/en/ from here
fast gate
fixed, thx
azure marshBOT
RVC Guides (How to Make AI Cover)
charred topaz
hello, I followed the website guide from the beggining, already created my database (clean sample) with UVR and then Audacity, I'm at the "Mainline" part (RVC), but I dont understand the "Inference" partm ut says "put your model's .pth inside the weights folder, but I dont know what a pth is, the zip RVS i downloaded have some .pth files
Im trying to create a voice model
the pth included in the "weights" file are:
guanguanV1
keruanV1
kikiV1
youzhanv2-xi
carmine hearth
If you are creating your own model, train the model with your own dataset first and then check the weights folder. The .pth files needed for inference are generated after training. Check the "Training" part first.
stable mauve
hi ive attempted to create a model based off of george strait.i stopped at 100 epochs however it sounded nothing like him.how can i go about making it better
thin stump
hi ive attempted to create a model based off of george strait.i stopped at 100 e...
Refer to the tensorboard. 100 epochs is not enough unless you are using the Ov2 pretraining
stable mauve
im using mangio
spring orchid
-colab
brittle wing
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
west agate
when training a model, if the audio is at 20khz do you train it at 40khz?
latent kettle
hello ... I have a problem.. I'm resuming the training but it is starting from 1 epoch. but I know last time it was 190 and I have all files including G and D, event, and rvclogs.zip please help me
brittle wing
Does anyone have any tips on how to use as few units as possible in Google Colab Pro
forest sluice
when training a model, if the audio is at 20khz do you train it at 40khz?
Works just fine
forest sluice
hello ... I have a problem.. I'm resuming the training but it is starting from 1...
on RVC Disconnected?
latent kettle
on RVC Disconnected?
yes
forest sluice
You clicked on the "Import Model from Drive to Notebook (for resuming)" cell right?
latent kettle
You clicked on the "Import Model from Drive to Notebook (for resuming)" cell rig...
yes I do everything
but not working
forest sluice
show me the logs folder
latent kettle
I don't know why this is happening.. everything I'm doing is correct
latent kettle
show me the `logs` folder
logs ?
forest sluice
Open the folder icon on the sidebar (left-hand side), open Mangio-RVC-Fork, open the logs folder, then the model name you use
latent kettle
Open the folder icon on the sidebar (left-hand side), open Mangio-RVC-Fork, open...
okay.. wait
latent kettle
Open the folder icon on the sidebar (left-hand side), open Mangio-RVC-Fork, open...
I don't know what to send but here it is
runtime disconnected.. but It have loaded G and D
latent kettle
this is what I'm getting on console screen
forest sluice
Well, it seems to resume from epoch 10
hallow ferry
<#1159290161683767298>
but that's RVC
sand iris
but that's RVC
The okada help channel was historically named "realtime"
... Because people were confused (they still are)
hallow ferry
oh
sand iris
Weird that it doesn't give any logs though
latent kettle
Well, it seems to resume from epoch 10
yeh it do... actually it was starting from 1 and I have trained it on 10 epochs and export it and started again to check that "will it continue from 10 or from 1"
azure marshBOT
RVC Guides (How to Make AI Cover)
pallid panther
what program can I use to make a cover? There are a lot of them. There is an ai voice changer, but it seems it just converts speech from the microphone and that's it
rare gobletBOT
Ayo? @pallid panther level 1 !!!
tribal igloo
carmine parrot
Does it happen to yall where discord just gets staticcy after a while? Its only discord tho
proper shale
https://colab.research.google.com/drive/1TU-kkQWVf-PLO_hSa2QCMZS1XF5xVHqs?usp=sh...
bruh what
just use rvc disconnected, check https://aihubdocs.github.io/en for guides
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
tribal igloo
is not Working
proper shale
is not Working
yeah, it is outdated
proper shale
just use rvc disconnected, check https://aihubdocs.github.io/en for guides
http...
refer to this message, please
velvet mulch
I am sorry I have a problem as well with continuing training my model.
I watched this Video https://www.youtube.com/watch?v=ixB9oalT3cQ This is also which I am using.
I am not sure how exactly I need to fill everything especially the Pretrained Paths . Did this already got solved somewhere or Someone got time to help me on this?
main lichen
with some models
i try to download
they have like 3 files
but i can only put in 2
so idk which one to put in
can someone help lol
proper shale
I am sorry I have a problem as well with continuing training my model.
I watched...
the pretrained paths should already be filled in
when you select sample rate and rvc version
proper shale
so idk which one to put in
what are the files that you have
proper shale
you only need the added index and the .pth file
weird that the model creator even included the TensorBoard logs lol
proper shale
just passes the normal, unchanged voice through
main lichen
Oh
rare gobletBOT
Ayo? @main lichen level 4 !!!
main lichen
Ok
One last one
Im speaking and on the app its says the volume is increased when i speak and thats normal
but i cant hear my playback on discord
proper shale
do you have the virtual audio cable already
if not, check https://rentry.co/VoiceChangerGuide
i have to go, so yea
main lichen
Yeah i have the vb cable
yeah bye
it was working fine but all of a sudden it just stopped working
low shard
I am sorry I have a problem as well with continuing training my model.
I watched...
that video is prolly outdated, its better if you check up our docs https://aihubdocs.github.io/en
also for help with w okada (ai voice changer) you can look in #1192011222023950368 and #🔍│help-w-okada
velvet mulch
that video is prolly outdated, its better if you check up our docs https://aihub...
thanks
low shard
yw
molten pecan
that video is prolly outdated, its better if you check up our docs https://aihub...
Nice job Nick ^-^
low shard
Nice job Nick ^-^
ty
valid spruce
Do you still have your training files like the folder of your model that was mad...
No
rare gobletBOT
Ayo? @valid spruce level 1 !!!
valid spruce
@odd shale Please teach me! I've been wondering how I can do this for a long time
@odd shale The Colab I used was RVC Disconnected
I tried training for 450 Epochse
glad zealot
Who made disconnected again? It was it kitlemon?
valid spruce
Yes
valid spruce
Who made disconnected again? It was it kitlemon?
And Noel Shirogane's
glad zealot
Kewk tanks
gleaming path
breh how do you send pictures here?
molten pecan
<@628603171735994379> Please teach me! I've been wondering how I can do this for...
https://aihubdocs.github.io/en/ here's a complete RVC Disconnected guide
odd shale
<@628603171735994379> The Colab I used was RVC Disconnected
Why you did ping me 2 times?
molten pecan
Then sorry, you can't retrain
In RVC-D yeah you can actually
odd shale
In RVC-D yeah you can actually
NO.
valid spruce
Why you did ping me 2 times?
Yes
odd shale
NO.
You can't retrain if you don't have your model folder that was created on the logs folder
And if you don't have your .pth that matches with the d_ and g_ checkpoint
valid spruce
Aquelas que tá escrito G_2333333 e D_2333333
odd shale
No, don't ping a staff member various times.
It's pretty annoying.
valid spruce
You can't retrain if you don't have your model folder that was created on the lo...
The ones that say G_2333333 and D_2333333
?
odd shale
The ones that say G_2333333 and D_2333333
If your G and D files have those name, then it means you mistakenly checked "Save the latest epoch" and "save an checkpoint everytime"
odd shale
The ones that say G_2333333 and D_2333333
Yes.
Also, even if you send screenshot of your weights folder, i'll won't be able to tell you from which .pth to continue training
valid spruce
If your G and D files have those name, then it means you mistakenly checked "Sav...
That's what I did
rare gobletBOT
Ayo? @valid spruce level 2 !!!
odd shale
That's what I did
Then sadly you won't be able to tell which .pth matches with your d and g files
Sorry.
gleaming path
bro if i try to connect the input of the microphone i always get this Looks like there's a bit of a problem.
unknwon message
If you clear the information being managed by this app, it may be recoverable.
valid spruce
Then sadly you won't be able to tell which .pth matches with your d and g files
So should I uncheck these boxes?
odd shale
So should I uncheck these boxes?
Only check one
valid spruce
Only check one
Which?
gleaming path
thx
rare gobletBOT
Ayo? @gleaming path level 1 !!!
odd shale
Which?
Only check one, if you want to save after finishing training or save an checkpoint everytime the training hits an amount of epochs.
valid spruce
Only check one, if you want to save after finishing training or save an checkpoi...
Oh
exotic citrus
Hello, I am new here, Ive downloaded a voice model.zip how do I use them now ?
odd shale
Hello, I am new here, Ive downloaded a voice model.zip how do I use them now ?
Check this guide.
exotic citrus
It tells from where to download the models but not how to use them, sorry Im confused
molten pecan
It tells from where to download the models but not how to use them, sorry Im con...
Check the AI cover page
=3
exotic citrus
Okay I found the page thank you ^^
valid spruce
Only check one, if you want to save after finishing training or save an checkpoi...
I think I understand, anyway, thank you very much for your help. ^^
main lichen
Bruhh
Theres like a one minute delay
but idk why
its on like 112 chunks
nvm
BRU what
It sometimes changes to a 5 second then 10 second then 3 second delays
cursive sedge
how ı can upload more voice on win 10?
restive ingot
How do I get .pht and .index files from a single .flac audio file?
terse bear
hello guys
anyone had this prob while training rvc
runtime\python.exe trainset_preprocess_pipeline_print.py Z:\AI\MYproject\samples 40000 7 Z:\AI\RVC-beta0717/logs/My vocal False
Traceback (most recent call last):
File "Z:\AI\RVC-beta0717\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: 'Z:\AI\RVC-beta0717/logs/My/preprocess.log'
gusty kestrelBOT
azure marshBOT
RVC Guides (How to Make AI Cover)
sand iris
hello guys
anyone had this prob while training rvc
runtime\python.exe trainset_...
spaces in filenames
deft dagger
can someone help me
brittle wing
Is it a problem if the path to the dataset folder is from Google drive?
gusty kestrelBOT
@brittle wing, I have found 2 results that match your search!
Requested by roses_are_blackpink
latent kettle
-colab
split latch
I am just gonna leave this here without sayin anythin 🙂
I kinda tried to remove sibliance and I think the sibliance on this one is better and not annoying
fluid matrix
oh god
dawn nebula
Is it a problem if the path to the dataset folder is from Google drive?
dont think so
fluid matrix
hello guys
anyone had this prob while training rvc
runtime\python.exe trainset_...
i dont know where you got that from but please delete that and follow the guide here https://aihubdocs.github.io/en/
terse shuttle
what is the best website to generate AI Covers? i already have a few voices in my google drive
the one i used a few months ago was taken down
spare shale
-colab
terse shuttle
which one is the most newbie friendly?
rare gobletBOT
Ayo? @terse shuttle level 1 !!!
terse shuttle
i just click on "install ilaria" and wait for a new window to pop up?
still totem
Ilaria RVC
Can it be used to train models?
odd shale
Can it be used to train models?
Nope, it can't be used for training models
still totem
What would you recommend for training?
odd shale
What would you recommend for training?
Mangio Fork, RVC Mainline or Applio
terse shuttle
Try using HF (HuggingFace)
error
i tried a different mp3 and gives me the same result
wait nvm, it worked
rare gobletBOT
Ayo? @dim snow level 1 !!!
odd acorn
Man my voice isnt working in RVC
odd shale
Man my voice isnt working in RVC
Your voice?
Or you wanted to say, voice model?
Or you're talking about W-Okada?
odd acorn
Yea
my voice isn't transmitting by RVC only other type of models, but yea its in W-okada
i think that i'm in the wrong chat ;-;
rare gobletBOT
Ayo? @odd acorn level 1 !!!
fluid matrix
Error: divison by zero
where did you get that error, can you show a screenshot or give a bit more context? what are you using?
half cove
split latch
gonna release some wild models soon bro
half cove
spare shale
@terse shuttle I Use EasyGUI To Train The Model
wanton meadow
can some one help me i have so much delay like 20k
rare gobletBOT
Ayo? @wanton meadow level 2 !!!
severe sand
What does this option do since we dont even have ckpt models?
half cove
What does this option do since we dont even have ckpt models?
for the training files
severe sand
Any point in using no?
does it allow you to restart training at an earlier epoch or something?
half cove
Can you tell me more on how to do that? Maybe I will need it some day
just delete the d and g files of the ones after you want to retrain at and retrain
severe sand
but I think those are saved regardless of the ckpt setting no? Or how does it relate to the setting
half cove
if you click that it will save them as d or g _ 2333333
severe sand
aah I see, thanks 👍
I canceled and restarted my training and now have multiple lines in tensorboard, can I remove the other ones somehow? the upper graphs are kinda funny tho ngl
odd shale
I canceled and restarted my training and now have multiple lines in tensorboard,...
No, you can't remove the other ones
You still got your training files?
Can't you start over again?
severe sand
Could some of you provide me with a short recording of your model translating pure "mmh" and "uuu" noises?
Every model I try is completely unable to do them (including my own) and I don't know if its because of my voice or an inability of okada.
I'd like to hear some examples of these sounds being produced properly.
broken oyster
guys i forgot what to do when applio decides to say hey you dont have torch so it just refuses to run
i already reinstalled everything including updating the dependencies
ruby lava
Use Ilaria RVC Colab
can i duplicate it??? cause i keep errors
azure marshBOT
RVC Guides (How to Make AI Cover)
brittle wing
where would i go to clone and train an ai voice
brittle wing
thanks
north torrent
im doing a voice cover of my friend singing some of alastors songs from hazbin hotel, how would i rid of the radio part
like the radio filter
thin stump
like the radio filter
that should be static/frequency noises you're referring to and denoising can get rid of that usually
molten pecan
it keep dong this
You are using a really outdated version
https://aihubdocs.github.io/en/ ← use the one explained here
That's also a usage guide
It's in RVC → Cloud → Inference
dawn notch
I tried to convert voice using rvc hf, but it will turn the voice into chipmunks😭 idk what's wrong, can someone help me, i'm trying to convert male voice to male not female, thankss😁
glad zealot
Set the pitch lower from what you set it to
molten pecan
I tried to convert voice using rvc hf, but it will turn the voice into chipmunks...
With Hugging Face? It's best if you use Colab instead. Much much faster & less buggy
https://aihubdocs.github.io/en/ check Ilaria RVC page
wicked trout
man i cant find any link to a collab where i can use any rvc model to change the voice of an audio
rare gobletBOT
Ayo? @wicked trout level 1 !!!
dawn notch
With Hugging Face? It's best if you use Colab instead. Much much faster & less b...
Okay let me try it, thankss
viscid moss
man i cant find any link to a collab where i can use any rvc model to change the...
-colab
idle gorge
anyone know where i can find the runtime area in github ive been searching for it for a while
junior halo
Okay, how is everyone usually making their song covers nowadays?? I've been using this colab link since June and the gradio thingy just now broke.
https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing&authuser=0#scrollTo=7vh6vphDwO0b
azure marshBOT
RVC Guides (How to Make AI Cover)
cedar cloud
@quick trench
quick trench
why not take a screenshot? you can send those on dms
What cell is that
and what link are you using...
this is the actual easyGUI colab ^
cedar cloud
Also had this problem
cedar cloud
https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw
Can only use for training models, not making covers
quick trench
Can only use for training models, not making covers
wrong.
it is for training, it is for inference
it's even for uploading models to huggingface if you want
- automatically saving things to google drive if you choose so
cedar cloud
I see
But using this to make another AI cover
The other link is the one I am having problems with
quick trench
Also had this problem
im the guy who made the easy gui. My colab is the real one
the other ones are copies, and so I can't really troubleshoot erros on their copies
lucid steppe
how to fix menu being blank when I need to select the model on so vits svc
fluid dust
Anyone know how to stop all the glitching and werid background sounds when using voice models
quick trench
Anyone know how to stop all the glitching and werid background sounds when using...
yess you need the highest quality audio, and good voice models
fluid dust
Is there a way to tell what is high quality
quick trench
listening to it, really
does it have noise apart from the voice? if it does, then it is not high quality audio
fluid dust
Yea there a lot of werid sounds in the back when I speak it’s good but when I’m not it’s making a punch of different noises
And does anyone have any preferred method they use to talk on discord or maybe live stream with the voice models
rare gobletBOT
Ayo? @fluid dust level 1 !!!
charred topaz
hello, i finally created a model and it worked fine, but now im creating another and this one just its really bad, and the time to create this new model is way faster, like way faster, the first one needed like 1.5 hours, the new one just minutes, im doing something diferent?
rare gobletBOT
Ayo? @charred topaz level 1 !!!
quick trench
Yea there a lot of werid sounds in the back when I speak it’s good but when I’m ...
yeah you need the cleanest audio if you want the cleanest ai results
quick trench
hello, i finally created a model and it worked fine, but now im creating another...
your dataset is probably smaller, am I correct? did you use less audio?
charred topaz
it is but just like 1 min shorter
quick trench
OHH you said it took minutes
that means you most likely put the wrong path to the dataset
so it didn't actually train anything
charred topaz
i think i know what happen, this one audio is like 10 mins and 10 seconds, and i selected the "Cache all training sets to GPU memory. Caching small datasets (less than 10 minutes) can speed up training,"
maybe is that?
charred topaz
ok, thanks, im gonna give it another try, like i said, my first model i created with some samples worked really well, im having the problem with this one im trying to create
quick trench
compare what you did different
is it wav files or a zip file? bc you cant train with a zip file for example
charred topaz
flac
quick trench
also make sure it doesn't have weird sybomls in the name of the file @#$ stuff like that
anddd most importantly
quick trench
that means you most likely put the wrong path to the dataset
verify it is the correct path
because i am 99% sure your error was that
I've had that error before, it happens when you accidentally put the wrong path
charred topaz
ok thanks im gonna give a nother try
quick trench
Okay
brittle wing
Can ilaria use datasets that we have trained
burnt ruin
this is what I get when i try to run step 3
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'
someone help please
quick trench
PYTORCH_PRETRAINED_BERT_CACHE = os.getenv("PYTORCH_PRETRAINED_BERT_CACHE", const...
Where did you get that colab? for the official easygui colab check my bio. The one you're using is broken.
odd cosmos
Hello, on "Step 2: Download the Model" I try pasting a link from the google excel sheet and it gives me an error, is there an updated excel sheet? none of the links work
thin stump
@thick oracle sorry for the ping. I just discovered how to link messages lol #✨│ai-help message
click that to see the thread
and scroll up for the context
sinful scroll
how do u find all the voices
deft dagger
is there any toturial on how to download the voice changer???
rare gobletBOT
Ayo? @deft dagger level 1 !!!
sinful scroll
is there any toturial on how to download the voice changer???
yeah check youtube
deft dagger
what do i search?? ai hub?
sinful scroll
what do i search?? ai hub?
Realtime AI Voice Changer Using RVC (Retrieval-based Voice Conversion w./ w-okada)
deft dagger
thanks
ember lily
Hi. Someone have problems with rvc? The public URL don’t run and I read so much code of error. I don’t need to train just running easy gui.
ember lily
PYTORCH_PRETRAINED_BERT_CACHE = os.getenv("PYTORCH_PRETRAINED_BERT_CACHE", const...
Same
latent kettle
-colab
glad zealot
Hi. Someone have problems with rvc? The public URL don’t run and I read so much ...
If you're using easy gui on the free version of colab, the gui will not work, it will run or a bit and just crash the whole runtime after a few minutes, you need to use the no gui version which is just on the top of the gui cell
elfin merlin
How do I even download it? I'm so confused, just joined
odd shale
How do I even download it? I'm so confused, just joined
What do you want to download?
proper shale
How do I even download it? I'm so confused, just joined
check https://aihubdocs.github.io/en for the voice conversion stuff
for the realtime voice changer: https://rentry.co/VoiceChangerGuide
azure marshBOT
RVC Guides (How to Make AI Cover)
full torrent
i have this error
Access denied with the following error:
Cannot retrieve the public link of the file. You may need to change
the permission to 'Anyone with the link', or have had many accesses.
You may still be able to access the file from the browser:
https://drive.google.com/uc?id=1t2C6tTYgxpc6fKXB0jMXv1etYcctPJPd
Error: Could not find a valid .pth file within the extracted zip.
If there's an error above this talking about 'Access denied', try one of the Alt URLs in the Google Sheets for this model.
what should i do?
please help me i want listen ai music with anime girl
odd shale
i have this error
Access denied with the following error:
Cannot retrieve ...
Probably the model stored in that drive link you used doesn't have an usable .pth file
odd shale
please help me i want listen ai music with anime girl
XD
I recommend you to use Huggingface links instead of Drive ones
keen cedar
Hello i see there is an experimental format shift interference, but when i clicked on it and tried to convert it said error
I would send a picture but its not letting me
finite galleon
Hello i see there is an experimental format shift interference, but when i click...
What was the error
keen cedar
just says error
keen cedar
llaria rvc
finite galleon
llaria rvc
Idk bout that
But formant shift is generally buggy
And really hard to control
keen cedar
oh wait
rare gobletBOT
Ayo? @keen cedar level 1 !!!
finite galleon
Plus it takes a long long time (10 minutes time for converting a 4 minutes audio)
finite galleon
You are using colab or na
keen cedar
on the cell?
finite galleon
on the cell?
Yes. You know the thing where you press run
keen cedar
im in the inference rn
finite galleon
Your using the web ui right
keen cedar
yes
finite galleon
Go to the colab page
keen cedar
@quick trench theirs
finite galleon
And see the output of the place where you pressed the button to run the web ui
finite galleon
Your on the colab page rn?
finite galleon
I know the tunnel that im using is Gradio
The part where you get the link is the output of the cell
There should be an error stated there
keen cedar
The part where you get the link is the output of the cell
AttributeError: 'NoneType' object has no attribute 'tobytes'
finite galleon
AttributeError: 'NoneType' object has no attribute 'tobytes'
Does normal inference work tho
Like without formant shift
keen cedar
yeah
finite galleon
I think you would need to ask @quick trench on that one
keen cedar
oh okay
finite galleon
As far as i know the no tobytes error is because of an invalid path
Which is not the case here
keen cedar
should I dm them or wait until i see them online
finite galleon
He kinda active rn so i say until he is online
subtle sable
Hello guys do You know how can i make to use an IA voice like Trump Mr beast Bad Bunny or from Hazbin from the channel to read any text i want? I want to use them in texts to talk but idk how pls someone tell me
fading summit
anyone have a new rvc google colab site?
rare gobletBOT
Ayo? @fading summit level 2 !!!
glad zealot
kaggle?
odd shale
Hello guys do You know how can i make to use an IA voice like Trump Mr beast Bad...
Usa Elevenlabs supongo
fading summit
rvc google colab doesnt work no more
forest sluice
rvc google colab doesnt work no more
What error?
fading summit
gradio and cloudflare doesnt work
which is step 3 doesnt work
which is weird cause it worked a few weeks ago
glad zealot
i use this one https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXI...
that one is outdated
fading summit
do u have a new one?@glad zealot
glad zealot
-colab
glad zealot
there
fading summit
which to use?
fading summit
ok thanks
zealous shuttle
anyone had this issue?
rare gobletBOT
Ayo? @brittle wing level 1 !!!
sweet quail
how can i get the zip file on hugging face
proper shale
-huggingface
azure marshBOT
-huggingface
proper shale
there it is
sweet quail
thanks
proud elk
Yo does anyone here have any tips for training AI as far as data sets and epochs go? I commonly see 250 epoch so im guessing thats some sort of magic number for this stuff
Also wondering how long sets should be or how much data you need or should have
brittle wing
everytime i reload to continue training this model, everything loads normally but the links don't show up even after waiting 20 mins?
timid haven
Easy GUI (for RVC v2, with crepe) (with improved downloader) is not working do you guys have alternative?
tidal cairn
everytime i reload to continue training this model, everything loads normally bu...
Same problem I’m having
rare gobletBOT
Ayo? @tidal cairn level 2 !!!
tidal cairn
If only someone would help us out this is super inconvenient
brittle wing
right
sweet quail
how can i remove my voice from the ai one because its mixed together with original voice and ai voice
proper shale
Easy GUI (for RVC v2, with crepe) (with improved downloader) is not working do y...
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
rvc disconnected, for training
if u want guides, https://aihubdocs.github.io/en
proper shale
Also wondering how long sets should be or how much data you need or should have
For a decent model, 5 mins would do (as long as you have good quality audio), but ofc I'd recommend anything from 10-30 mins. Also, for epochs, it really varies. You have to check the graphs and see if it ain't overtraining
proper shale
where those at?
on TensorBoard, it depends on what you're using - if you're on Colab it's already there
glad zealot
For a decent model, 5 mins would do (as long as you have good quality audio), bu...
me using 2 hours for no reason lmao
proper shale
me using 2 hours for no reason lmao
😭
proud elk
MB im using mangio rvc web
rare gobletBOT
Ayo? @proud elk level 1 !!!
proud elk
different toolset im guessing
glad zealot
nah its the same
proper shale
MB im using mangio rvc web
oh, local user!
https://aihubdocs.github.io/en/rvc/local/mainline-rvc/ has a part explaining how to install TensorBoard, this applies to all forks
glad zealot
the only different from mainline i think is the ability to stop/pause traning midway
sweet quail
is easy gui rvc gonna get fixed?
glad zealot
its not showing a link to gradio
ye it wont unless you are on colab pro
it depends if the creator of it wants to add it back
sweet quail
ye it wont unless you are on colab pro
this?
rare gobletBOT
Ayo? @sweet quail level 2 !!!
glad zealot
since the method of bringing it back is kinds inthe grey area in my opinion
modest isle
why doesnt my mic work for w okada?
sweet quail
since the method of bringing it back is kinds inthe grey area in my opinion
i tried the hina colab but my own voice is mixed into ai how do i remove that kind of effect
glad zealot
-colab
glad zealot
https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sh...
yup seems outdate
d
lmao
sweet quail
yup seems outdate
is there new version?
glad zealot
<https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw>
this is the new one
glad zealot
i tried the hina colab but my own voice is mixed into ai how do i remove that ki...
voice changer or covergen?
sweet quail
voice changer or covergen?
voice changer
its my voice and ai combined in 1
idk how ill removed my own voice to the ai one
glad zealot
can you screenshot the setting you have?
sweet quail
can you screenshot the setting you have?
alr gimme a sec i got diconnected
glad zealot
also prob better to ask about voice changer in #🔍│help-w-okada
worldly hound
on my computer MDX-NET Colab is not working. Can someone help me?
half cove
yea that colab is pooped rn
glad zealot
on my computer MDX-NET Colab is not working. Can someone help me?
i think thats the old one which doesnt get updated anymore
glad zealot
yea that colab is pooped rn
me with a working one not sure if they are gonna update or fix that one tho
half cove
~~me with a working one~~ not sure if they are gonna update or fix that one tho
i tried fixin it but too lazy
worldly hound
i think thats the old one which doesnt get updated anymore
OK thank you
half cove
give em the fixed one 🗣️
glad zealot
lemme try it now if i can fix it
btw did google fix the cuda problem or still need to do it outselves?
brittle wing
if a chorus part is lower than the lead vocal, is rmvpe still get the pitch correctly?
proper shale
ehhh I wouldn't bet on it, I'd rather filter that out
worldly hound
lemme try it now if i can fix it
do you know any other vocal remover that removes 100% of the instrumental
glad zealot
do you know any other vocal remover that removes 100% of the instrumental
msvep?
glad zealot
for me it just always used the wrong version of numpy
wtf np.complex is not even used anywhere
half cove
wtf np.complex is not even used anywhere
old version of librosa or somthin
if you update it then that error goes away
but same error happens in second cell
glad zealot
kewk lemme try
storm holly
<https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw>
how do i use this one?
for covers not model making
glad zealot
for covers you can use either covergen or ilaria rvc
brittle wing
bro input problem please help
half cove
btw rejekts should have made it so the gui works on free accounts so test it on a alt and say if it works
low shard
btw rejekts should have made it so the gui works on free accounts so test it on ...
Yea
Its easy bypass with rot 13
half cove
Its easy bypass with rot 13 <:lfg:1159355870119993496>
he just changed app.py to demo
and it works
low shard
Good
glad zealot
HAH NO ERROR
junior halo
I am so confused. I am in the gradio page of ther Hina Mod colab project. When I go to generate it, it's saying there is no nvidia GPU installed? I thought my system didn't matter and all of this is processed entirely on the web rather than using my resources?? It won't let me generate the final output without an nvidia GPU. Makes no sense.
It was working fine yesterday.
glad zealot
are you connected with the t4 gpu?
junior halo
OHH
half cove
if it doesnt say t4 make sure to change it to t4 runtime
junior halo
It does not.
It only has the option to connect without GPU.
Does that mean I gotta pay for more GPU usage?
half cove
It only has the option to connect without GPU.
click the down arrow then changeruntime type
glad zealot
i tried fixin it but too lazy
kekekke
half cove
junior halo
I can't change it to T4.
rare gobletBOT
Ayo? @junior halo level 3 !!!
glad zealot
fixed the cuda thing too since it wont use gpu
half cove
I can't change it to T4.
what does it show
glad zealot
I can't change it to T4.
switch account or try again tomorrow
half cove
switch account or try again tomorrow
i mean it should still show up if you ran out of runtime or am i wrong?
glad zealot
ye it will say youre out of usage or something
glad zealot
i havent even used colab what <:skullsob:1159372531992645662>
lmao
glad zealot
on my computer MDX-NET Colab is not working. Can someone help me?
try this link
aight gn that should work
worldly hound
aight gn that ***should*** work
Same problem, I think I'll use Ilaria uvr
glad zealot
Same problem, I think I'll use Ilaria uvr
nope works fine to me, you pressed the link i just sent right?
kindred bloom
worldly hound
nope works fine to me, you pressed the link i just sent right?
yes
worldly hound
maybe it's my computer
half cove
waht did i do
half cove
anyways ye imma go sleep
gn 
glad zealot
waht did i do <:boohooh:1176674698629750975>
¯_(ツ)_/¯
worldly hound
It doesn't work, I've already tried 5 times, even on my cell phone, I'm going to create another account on the drive to see if it will work.
rare gobletBOT
Ayo? @worldly hound level 4 !!!
odd shale
It doesn't work, I've already tried 5 times, even on my cell phone, I'm going to...
Nah, that colab is broken
half cove
Nah, that colab is broken
brittle wing
where is the folder "Models" and "Tracks" to use the collab?
rare gobletBOT
Ayo? @brittle wing level 1 !!!
glad zealot
dw it works
finally i can close the 20 tabs i used to fix that XD
half cove
sharp pawn
sharp pawn
Bro screenshoted his own message <:skullsob:1159372531992645662>
didnt wanna type it up again
sharp pawn
Look, u must execute the next cells in order.
also, im asking about the place where i upload my sample
like my wav that i wanna train
sharp pawn
doesnt matter
azure marshBOT
RVC Guides (How to Make AI Cover)
cerulean rover
assuming I use a voice model from here in youtube videos, besides giving credit to the author of the model, do I have to do something else?
half cove
assuming I use a voice model from here in youtube videos, besides giving credit ...
just crediting should be fine
odd shale
assuming I use a voice model from here in youtube videos, besides giving credit ...
Just credit, but don't do it if the OG author doesn't want. (Likely to avoid problems with the OG VA of the character)
wild wraith
i have 11 minutes voice record how many epoch should i choose
odd shale
i have 11 minutes voice record how many epoch should i choose
From 300 to 500
wild wraith
From 300 to 500
thx a lot
crude flame
Am i wrong in saying this model mode collapsed?
odd shale
Am i wrong in saying this model mode collapsed?
Yeah, i think it collapsed pretty hard
Maybe your dataset wasn't clean enough or you used an excessively big batch size
crude flame
I used a batch size of 8
rare gobletBOT
Ayo? @brittle wing level 2 !!!
odd shale
I used a batch size of 8
Which was your dataset length?
crude flame
Just under 15 minutes
odd shale
Just under 15 minutes
I think you could try with a different batch size.
modern oxide
charred topaz
hello, the first time i created my first model (trained it) the process was slow and that model worked really well, now im trying to create another voice model but this this the "Epoch" process really fast
like about 1 second after another
and I remember that first model i created it needed way more time between the epochs
odd shale
and I remember that first model i created it needed way more time between the ep...
The time between epoch and epoch will depend on your dataset length and your batch size.
Which batch size you used and which was your dataset length?
charred topaz
dataset is about 35 min long, and the "Batch size per GPU" i used was 5
i dont even know what Batch size per GPU is i just let it on default 5
brittle wing
Hi @molten pecan I can't really get TensorVenv to work. When I open it, it says the site can't be reached
charred topaz
i just created the new model and sounds the same as the one i created before (not my first one), just a bad
molten pecan
Hi <@791013455251374151> I can't really get TensorVenv to work. When I open it, ...
In which version of RVC?
brittle wing
@molten pecan Uhh how do I see that? I'm very new to this so I don't know all this stuff. Do you mean like what it's called or? BEcause If so I believe Mangio-RVC v23.7.0
It didn't come with a tensorfile, but another tutorial on the website told me to put the tensor file into the RVC folder
charred topaz
i think this discord is for the RVC WebUI
brittle wing
Uhm, sorry mate i don't know what that is. I found tutorials and links and stuff for Mangio on a link given on this server
So I assumed that I could get help for that here
But am I doing it wrong? I really have no idea how to do this, I can make covers and stuff, I just don't know how to get those damned files (PTH and Index) For my voice model
If I could get those I can do the song making and stuff I have done that with downloaded models. But I wanna make one of my own voice
odd shale
dataset is about 35 min long, and the "Batch size per GPU" i used was 5
Do you have Yeah.. I think you should use a higher batch size.
dawn nebula
@brittle wing probs install mainline rvc or applio
charred topaz
Do you have Yeah.. I think you should use a higher batch size.
how it works? I have a RTX 3080
rare gobletBOT
Ayo? @charred topaz level 2 !!!
odd shale
how it works? I have a RTX 3080
Watch if your PC handles a higher batch size.
dawn nebula
how it works? I have a RTX 3080
just set it to 8, probs reduce dataset length a bit coz rvc sucks with long dataset
odd shale
just set it to 8, probs reduce dataset length a bit coz rvc sucks with long data...
Delik, he already had done that.
dawn nebula
oh
odd shale
how it works? I have a RTX 3080
Or instead of using a higher batch size, cut the dataset to 24 mins.
dawn nebula
he use 5?
proud elk
I am trying to train a model I previously worked on but I keep getting a runtime error: RuntimeError: The expanded size of the tensor (12800) must match the existing size (0) at non-singleton dimension 1. Target sizes: [1, 12800]. Tensor sizes: [0]
charred topaz
i think i screw up at the sample rate
proud elk
any reason?
dawn nebula
any reason?
is this an rvc v1 model
proud elk
nope
dawn nebula
show me the logs/<experiment name> folder
dawn nebula
how many epochs did u train
proud elk
just 20
as a start
could always just redo it tbh
maybe I messed it up by re-doing the whole process
is it possible to just skip to step three on a re-train?
dawn nebula
just retrain
charred topaz
ahh I think I know what i did bad, i my files are 44100 sample rate and in teh rvc I put 48k?????
is that?
odd shale
just retrain
A dataset of 25 mins is fine?
dawn nebula
Am i wrong in saying this model mode collapsed?
think so, dataset issue
proud elk
I did a 45ish min dataset and 100 epochs worked wonderfully for me
sounded quite natural
dawn nebula
I did a 45ish min dataset and 100 epochs worked wonderfully for me
its because ur dataset is good
proud elk
Yeah it was literally my girlfriend reading harry potter lmao
charred topaz
the TensorBoard doesnt work for me, it says cant finde the folder name in cmd and doesnt load anything after that
my Pc username has 2 words maybe is cuz that?
dawn nebula
does it have a space in between?
charred topaz
yes