What should i do
#✨│ai-help
soft stratus
glad zealot
If it's your model that you made you can upload it on huggingface
astral citrus
When trying to export a RVC as ONNX, It keeps failing and saying that I dont have a NVIDIA Graphics Driver installed. But I set everything up for AMD
rare gobletBOT
Ayo? @astral citrus level 1 !!! 
astral citrus
Also ============= Diagnostic Run torch.onnx.export version 2.0.1+cu118 =============
verbose: False, log level: Level.ERROR
======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ========================
Also that I have no space on my drive even though I have 250 gb left
glad zealot
Cant help it's local sorry
astral citrus
Rip, I recently moved it from my C drive to my other drive. Do you think it could've fucked up there?
glad zealot
Dang it idk who to ping
I don't think directory change should do anything to it
As long as the directory doesn't have spaces
astral citrus
Its weird, it works sometimes and not others
but even the .onnx file doesnt work when uploading it
Yea weird I think it keeps trying to DL to my C drive
File "torch\onnx_internal\onnx_proto_utils.py", line 175, in _export_file
OSError: [Errno 28] No space left on device
ima just reinstall ty though LOL
glad zealot
It should say finished training
brittle wing
where writed processing
?
8 minute data set has been training for a long time
I put a dataset 8 minutes long without cuts. He'll take care of cutting it, right? @glad zealot
glad zealot
Yes
brittle wing
@glad zealot i don't find index file
I don't understand why he created so many models for me
But in the log folder, I can't find the index
(in "papa" folder
glad zealot
You need it press index to get index
brittle wing
You need it press index to get index
Where
glad zealot
In the training also
glad zealot
No index then
The model will work without index anyways
So not really that important
brittle wing
So not really that important
If I don't select that the audio it generates will be normal @glad zealot
When I generate the voice it's not that. I don't understand, should I start the training all over again?
glad zealot
Try just putting the folder not the audio file itself
brittle wing
Try just putting the folder not the audio file itself
where
rare gobletBOT
Ayo? @brittle wing level 5 !!!
loud pond
Does anyone have VC on mac? I am so confused on how to install
brittle wing
Try just putting the folder not the audio file itself
I don't understand what you mean where should I put the folder?
glad zealot
Put the path of the folder
brittle wing
Put the path of the folder
the voice is different ;((((
glad zealot
Just make a new folder for each voice
glad zealot
Oh ye you can use any index
brittle wing
Just make a new folder for each voice
In short, a different voice comes out, a voice not created and unknown by me
violet heron
I don't understand why it created all these files for me
That’s your model you trained
The e stands for epochs and are checkpoints
violet heron
The e stands for epochs and are checkpoints
Basically saves of the model when training
brittle wing
Index is for accent
How can I listen to the .path file to understand if it has a good voice. this voice is different and unknown
violet heron
How can I listen to the .path file to understand if it has a good voice. this vo...
Go to inference
Select the model
Put the path to your audio file
Generate/Convert/Create button
Wait a minute there’s your cover
brittle wing
Go to inference
Understand?
I select my model file .pth
This is not my voice
its random voice
violet heron
its random voice
So the model you trained doesn’t sound like the dataset?
brittle wing
So the model you trained doesn’t sound like the dataset?
Correct
brittle wing
In my folder
violet heron
Try a different audio to convert
Or change the transpose if the voice too high or low
brittle wing
Try a different audio to convert
Work difference audio
brittle wing
Or change the transpose if the voice too high or low
I've tried with many
Different
violet heron
Work difference audio
So it doesn’t work on that audio but it works on others
brittle wing
So it doesn’t work on that audio but it works on others
Correct
How can I listen to the .pth file to understand if it has a good voice. this voice is different and unknown
?
brittle wing
The audio or the model
I have to tow it all over again
late wave
did you find a solution? im having the same problem
violet heron
The audio or the model
Because it just might be a bad song
dusty sparrow
Does anyone know if something changed about RVC v2 google colab or google drive? I suddenly cant download models fromy my own drive despite the permissions being set correctly. Just worked for me yesterday
violet heron
Does anyone know if something changed about RVC v2 google colab or google drive?...
Can you link the colab
dusty sparrow
brittle wing
When I open I Mangio RVC. Select papa.pth
I press Convert and an unknown voice comes out. An audio with different voice from the model and dataset
violet heron
https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sh...
Very outdated
brittle wing
I don't have the index file. He didn't generate it for me
violet heron
Very outdated
-colab use one of these
azure marshBOT
-colab use one of these
dusty sparrow
figures lol I haven't been checking the discord like I should be
brittle wing
-colab use one of these
I don't have the index file. He didn't generate it for me
brittle wing
The audio or the model
When I open I Mangio RVC. Select papa.pth
I press Convert and an unknown voice comes out. An audio with different voice from the model and dataset
violet heron
I don't have the index file. He didn't generate it for me
Enter your dataset and the same information in the train tab, don’t train model but click train feature index
violet heron
When I open I Mangio RVC. Select papa.pth
I press Convert and an unknown voice ...
When you convert a different audio it sounds like the dataset?
If so that’s probably the songs fault
brittle wing
Enter your dataset and the same information in the train tab, don’t train model ...
I don't understand
violet heron
I don't understand
Enter the same info you used to train your model with
Like do it again
But
Instead of clicking “Train Model”
brittle wing
All models sound as they should. Only this new model. It has an unknown voice for all conversions and with all music
violet heron
Instead of clicking “Train Model”
You click “Train Feature Index”
violet heron
All models sound as they should. Only this new model. It has an unknown voice fo...
Oh
So it’s the model
brittle wing
I don't understand why it generated so many pth files for me
violet heron
Batch size sometimes is the reason
Sometimes it’s sampling rate
Sometimes it’s overtraining/undertraining
Might just be because you don’t have the .index file
brittle wing
I started train with these settings
violet heron
Might just be because you don’t have the .index file
Which is basically the accent
violet heron
Which is basically the accent
So the fact you don’t have a index file is most likely the reason
violet heron
Train again with new settings?
Train again same settings but only train the index
Because you already have the model
brittle wing
If it takes time I'll do everything together now since I'm going to sleep
brittle wing
Train again same settings but only train the index
If train only index is fast then I do it
young halo
Does someone know what is artifacting and how can I fix it?
brittle wing
Train again same settings but only train the index
Its correct here? Is there some error here? @violet heron
rigid pelican
No matter what voice I'm using, the RVC says the voice is downloaded, but when I refresh it only says Jake. No matter what voice I'm linking it's just Jake. No other option.
I don't even know who Jake is.
Jake the dog?
brittle wing
violet heron
For train feature index?
violet heron
If train only index is fast then I do it
Expected
Now you have your index file
Try converting something using the same model with that index
brittle wing
Now you have your index file
I've been train from the start
brittle wing
with colab
warm fractal
1660 super and i5 3600x
violet heron
I've been train from the start
Ok
brittle wing
See some problems here?
sour fern
Does anyone know if something changed about RVC v2 google colab or google drive?...
I'm having the exact same problem suddenly with this same link - any luck yet?
rigid pelican
@sour fern It said mine downloaded but it doesn't show up when I refresh. The only thing listed is Jake
dusty sparrow
<@1071894258556088410> It said mine downloaded but it doesn't show up when I ref...
Seems like this is a google drive issue more than colab or it's both. The program runs fine with the huggingface link. I bet the models would download properly with the colab if you used a huggingface link but I haven't tried it. glad to see it's not only just me. i pay for colab subscription so kinda frustrating lol
rare gobletBOT
Ayo? @dusty sparrow level 1 !!!
brittle wing
start preprocess
['trainset_preprocess_pipeline_print.py', 'C:\Users\Vincenzo\Downloads\data\dataset.wav', '48000', '4', 'C:\Users\Vincenzo\Desktop\CoverAInew\Mangio-RVC-v23.7.0/logs/papaa', 'False']
Fail. Traceback (most recent call last):
File "C:\Users\Vincenzo\Desktop\CoverAInew\Mangio-RVC-v23.7.0\trainset_preprocess_pipeline_print.py", line 118, in pipeline_mp_inp_dir
for idx, name in enumerate(sorted(list(os.listdir(inp_root))))
NotADirectoryError: [WinError 267] The directory name is invalid: 'C:\Users\Vincenzo\Downloads\data\dataset.wav'
end preprocess
rigid pelican
Seems like this is a google drive issue more than colab or it's both. The progra...
But it IS the huggingface link. I just downloaded it and put it in my google drive to better categorize the voice models by character and show.
rare gobletBOT
Ayo? @rigid pelican level 1 !!!
rigid pelican
@dusty sparrow Plus, It says Downloaded after I put the link and Name. But it won't show up in the list when I refresh.
sour fern
Seems like this is a google drive issue more than colab or it's both. The progra...
Agreed, this is only affects my models stored in Google Drive. I don't know anything about Huggingface other than seeing the link everywhere, I guess I could move my models over there if necessary? Like you, I already pay for Colab, so...
modern surge
-colab
violet heron
But it IS the huggingface link. I just downloaded it and put it in my google dri...
Then that makes it a google drive link
brittle wing
start preprocess
['trainset_preprocess_pipeline_print.py', 'C:\Users\Vincenzo\Downloads\data\dataset.wav', '48000', '4', 'C:\Users\Vincenzo\Desktop\CoverAInew\Mangio-RVC-v23.7.0/logs/papaa', 'False']
Fail. Traceback (most recent call last):
File "C:\Users\Vincenzo\Desktop\CoverAInew\Mangio-RVC-v23.7.0\trainset_preprocess_pipeline_print.py", line 118, in pipeline_mp_inp_dir
for idx, name in enumerate(sorted(list(os.listdir(inp_root))))
NotADirectoryError: [WinError 267] The directory name is invalid: 'C:\Users\Vincenzo\Downloads\data\dataset.wav'
end preprocess
this is error
whyyy
rigid pelican
<a:skulld:1148612375302385676>
So we can only use direct huggingface links now? Despite the fact that when I go to the Voice Models category and look for voices, I only find download links that say Huggingface but no links to the suppose huggingface site or whatevere it is?
brittle wing
<a:skulld:1148612375302385676>
How can I install RVC Mangia ect locally?
The link??
rigid pelican
@violet heron let's assume I know nothing beyond a bunch of useless art program jargon. Does this mean that I downloaded all these voices I downloaded and uploaded to my Google Drive are completely useless now?
brittle wing
-rvc
rare gobletBOT
Ayo? @brittle wing level 6 !!!
sour fern
<@858923165379199007> let's assume I know nothing beyond a bunch of useless art ...
This is the problem I'm working on now - I'm hoping it's fixable and not Google actively doing something to prevent these models from working from Drive. But like you, I am only a useless artist.
rare gobletBOT
Ayo? @sour fern level 1 !!!
rigid pelican
@sour fern I'm gonna cry. I can't get one piece of technology to work with me ever. Thank you anyway.
finite galleon
This is the problem I'm working on now - I'm hoping it's fixable and not Google ...
That might be the issue google is trying to prevent it
violet heron
<@858923165379199007> let's assume I know nothing beyond a bunch of useless art ...
You can still use em
violet heron
You can still use em
Some HF spaces support them
finite galleon
Last week it works
violet heron
And google colabs do
finite galleon
You can still use em
Currently no
modern surge
Why does this error happen when I paste the link?
finite galleon
Google drive downloads are broken rn
rigid pelican
@violet heron The advanced RVC says it supports Google Drive links, but when i try to upload it it says error
finite galleon
Why does this error happen when I paste the link?
Yeah this is the reason cannot retrieve file
Dont use a drive link
finite galleon
<@858923165379199007> The advanced RVC says it supports Google Drive links, but ...
Currently its broken
modern surge
Damn, I have to upload 5 models right now.
violet heron
<@858923165379199007> The advanced RVC says it supports Google Drive links, but ...
Guess it doesn’t work any more 
Yeah your going to have to upload them to HuggingFace
If your using HuggingFace Spaces at least
finite galleon
Blame google for having a virus scan now
modern surge
How do I upload my model to Hugginface?
finite galleon
How do I upload my model to Hugginface?
-huggingface
azure marshBOT
-huggingface
Suggestions for @modern surge
modern surge
-huggingface
modern surge
Show me the input
Can i change the repository to public?
rare gobletBOT
Ayo? @modern surge level 5 !!!
finite galleon
Can i change the repository to public?
Ye
finite galleon
Okay, still the same can't be downloaded
Your link is a HuggingFace link right
modern surge
It tells me that the file is not a zip now
finite galleon
If its not a huggingface link it wont work
modern surge
It's a zip.
modern surge
If its not a huggingface link it wont work
It is
finite galleon
https://huggingface.co/Geniuhh/Gencalibre/blob/main/gencalibre.zip
Is it zipped correctly
rare gobletBOT
Ayo? @finite galleon level 27 !!!
violet heron
https://huggingface.co/Geniuhh/Gencalibre/blob/main/gencalibre.zip
Change blob to resolve
In the link
modern surge
finite galleon
Change blob to resolve
Ohe ye i forgor
modern surge
Change blob to resolve
I remove it from the link or do I have to do something else?
modern surge
Ohe ye i forgor
How?
fossil vigil
i got this error that says "traceback (most recent call last): File " directory of rvc" line 129 in vc_single
File "rvc directory/vc infer pipeline line 340 in pipeline
what do i do
and says some more stuff
any way i can send the log
violet heron
I remove it from the link or do I have to do something else?
Replace blob to resolve
On the link
violet heron
Can anyone tell me how to make the voice sound human?
Use RVMPE instead of crepe
sharp parcel
Use RVMPE instead of crepe
thats what i did
rare gobletBOT
Ayo? @sharp parcel level 2 !!!
sharp parcel
but its till robotic
fluid thicket
but its till robotic
There is also pitch to change such as if its female led song initially you need to -12 for male models and for the opposite, male led vocals and female mode +12 (though if you encounter issue of it pitching out so high then back down to like +8)
rare gobletBOT
Ayo? @fluid thicket level 1 !!!
sharp parcel
There is also pitch to change such as if its female led song initially you need ...
I used 5 and it was fine
But the vocals don't match the ups and down of the instruments
It feels monotonous
Like a robot
fluid thicket
Kind of like its slowed down you mean?
Because yeah if I pitch it outside of the -12 / 0 / +12 range I do encounter that
sharp parcel
Kind of like its slowed down you mean?
No
Imagine your just narrating the lyric instead of singing
When there is a song
fluid thicket
Can you drop a lil sample of the worst part in here?
wind geyser
Anyone know how to fix the issue of where like, your talking and you play it back, the voice is low then all of a sudden it gets loud
sharp parcel
Can you drop a lil sample of the worst part in here?
Can I dm you?
fluid thicket
Just post it here because if I don't know someone may come along and have the answer
sharp parcel
Just post it here because if I don't know someone may come along and have the an...
We are not allowed to post here
That's what I was told by the mods
azure marshBOT
RVC Guides (How to Make AI Cover)
fluid thicket
I could be wrong but I think that is just for copyright material. Just posting the vocals only no instrumental should be fine. Kind of like when people post sample audios of their models
vast berry
Which RVC colab working right now?
hallow wren
Hey @molten pecan , hope you are well. I was using your space as duplicate one but it's currently not working. Is anything wrong ?
sharp parcel
I could be wrong but I think that is just for copyright material. Just posting t...
Well I don't want to take a chance
vast berry
Working?
vast berry
Yee
cinder matrix
Кто может пойти в лс и помочь с vs client
brittle wing
What are the new updated files for rvc mangio and ect in local? I currently only have pm god and harvest @acoustic thorn
eager pelican
the pest pitch extraction is ur-mom
RMVPE
zealous shuttle
need best setting for cleanest acapella
brittle wing
['trainset_preprocess_pipeline_print.py', 'C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\datasets\dataset.wav', '48000', '4', 'C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0/logs/papa', 'False']
Fail. Traceback (most recent call last):
File "C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\trainset_preprocess_pipeline_print.py", line 118, in pipeline_mp_inp_dir
for idx, name in enumerate(sorted(list(os.listdir(inp_root))))
NotADirectoryError: [WinError 267] The directory name is invalid: 'C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\datasets\dataset.wav
Why this error guys?
sand iris
['trainset_preprocess_pipeline_print.py', 'C:\Users\Vincenzo\Desktop\AICOVER\Man...
it asks for a folder
you gave it a file
try with C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\datasets\
brittle wing
the dataset.wav?
brittle wing
it asks for a folder
If I select only the folder will it take care of getting the wav file?
C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\datasets\
No C:\Users\Vincenzo\Desktop\AICOVER\Mangio\Mangio-RVC-v23.7.0\datasets\dataset.wav ?
zinc anchor
Looks like it
azure marshBOT
RVC Guides (How to Make AI Cover)
forest sluice
Working?
A better way, use Windows+Shift+S, then drag a portion of your screen, then paste it here
crimson wraith
is there a better rvc tts than applio? because with applio i have to use a tts model from edge or google which isn't giving decent results, my character has an accent btw so that's one of the reasons why
finite galleon
is there a better rvc tts than applio? because with applio i have to use a tts m...
If your looking to only use tts, rvc might not be a great choice
You would need to train a model that isnt rvc like using ElevenLabs (paid), TortoiseTTS, StyleTTS, etc using the the audio that you want to clone
crimson wraith
If your looking to only use tts, rvc might not be a great choice
not entirely, just using it for a thing i'm doing rq, but i mostly use it for covers, but i just wanted to know if there are better alternatives for rvc tts
thanks though!
gritty socket
How do I verify?
west cosmos
Is anybody else suddenly having issues sharing models on drive to hugging space? I've tried numerous spaces, different google accounts, different models etc.
In the logs it seems to always through this error message up "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. "
The models all have the same sharing permissions they did a few weeks ago so no change there
Apologies I've just seen Eempostor mentioned drive links aren't working right now. I'll see about uploading elsewhere.
jolly fable
Hi, is there a guide to installing tensorboard? I found a guide but the github that links for installation is no longer existing
rare gobletBOT
Ayo? @jolly fable level 1 !!!
proper shale
_ _
fossil vigil
can someone help me? the converting process wont work and i end up with this error
proper shale
@jolly fable
proper shale
can someone help me? the converting process wont work and i end up with this err...
wait what's your GPU
fossil vigil
prolly a pc problem since my computer is as old as me
proper shale
Maybe try a more recent local fork
jolly fable
wait what's your GPU
2070s
proper shale
-local
azure marshBOT
fossil vigil
will this work on a gtx 1060
kind ether
what does speaker id do?
proper shale
will this work on a gtx 1060
Yeah...
proper shale
what does speaker id do?
It was a feature that never really came out, so don't touch it
forest sluice
can someone help me? the converting process wont work and i end up with this err...
Did you do either of these commands:
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
or
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
To install Pytorch (or is your downloaded RVC the prepackaged?)
jolly fable
any help to understand how to choose the correct .pth in the fodler? They all sound quite similar
proper shale
any help to understand how to choose the correct .pth in the fodler? They all so...
The PTH you're looking for is in weights, and check for the most recent one
fossil vigil
Did you do either of these commands:
```
pip3 install torch torchvision torchaud...
I’m not sure, but I think it’s a pc error bc it just bluescreened
I’m on my phone rn
I’ll borrow my siblings laptop
forest sluice
Did you do either of these commands:
```
pip3 install torch torchvision torchaud...
There's also a similar issue that has been resolved in a1111 repo https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/8965#issuecomment-1530085758 maybe have a look at it
fossil vigil
Yeah it’s a pc error
rare gobletBOT
Ayo? @fossil vigil level 2 !!!
fossil vigil
Ayo? <@860918415137308692> level **2** !!! <:lfg:1159355870119993496>
Fact
jolly fable
The PTH you're looking for is in weights, and check for the most recent one
ty, they look like this, in the guide I was following it said to check with the tensorboards which was the best among the various 'saves', doesn't that make sense?
proper shale
ty, they look like this, in the guide I was following it said to check with the ...
yeah, it makes sense, but try the latest one either way
forest sluice
ty, they look like this, in the guide I was following it said to check with the ...
alternatively, test all models
jolly fable
alternatively, test all models
lol 101 models 😄
jolly fable
yeah, it makes sense, but try the latest one either way
I tried first, in the middle and last, the middle one "seems" to sound better
forest sluice
If it's better, use it. If you want to upload, select your preferred model + index file into a folder, then zip it
-upload
jolly fable
If it's better, use it. If you want to upload, select your preferred model + ind...
I still check a bit if there is a better one, can I upload it if it is a singer?
molten pecan
Hey <@791013455251374151> , hope you are well. I was using your space as duplica...
hehe I didn't make the space, @proven hill did. but what's the matter?
you can send a screenshot
forest sluice
I still check a bit if there is a better one, can I upload it if it is a singer?
completely fine, as long as your sample audio doesn't contain instrumentals
jolly fable
completely fine, as long as your sample audio doesn't contain instrumentals
great, ty
molten pecan
completely fine, as long as your sample audio doesn't contain instrumentals
i just realized
your pfp is gura?
jolly fable
is there a channel on which to upload audios for them to be 'evaluated'?
molten pecan
is there a channel on which to upload audios for them to be 'evaluated'?
uhm not really. we evaluate models and the audio samples, but not exclusively audios
jolly fable
uhm not really. we evaluate models and the audio samples, but not exclusively au...
I found the italian chat maybe there
molten pecan
we ask to include audio samples that represent their model. but the goal is to know if the model is good or not
jolly fable
just to ask "how similar"
molten pecan
I found the italian chat maybe there
make sure it's not copyrighted
jolly fable
make sure it's not copyrighted
ok
spare solstice
guys how and where do i upload the model so it can be accessed by others
proper shale
-huggingface
azure marshBOT
-huggingface
proper shale
On Huggingface
And then use QCBot to try and get model maker in #1159514067187277865
Then if you're approved post on #1175430844685484042
molten pecan
guys how and where do i upload the model so it can be accessed by others
.
there's your answer
spare solstice
thank you both of you
proper shale
you're welcome:)
stuck bane
anyone know how to fix??
RuntimeError: Expected tensor for argument #1 'input' to have the same type as tensor for argument #2 'weight'; but type torch.cuda.FloatTensor does not equal torch.cuda.HalfTensor (while checking arguments for cudnn_batch_norm)
@modern oxide@peak tusk
rare gobletBOT
Ayo? @stuck bane level 7 !!!
modern oxide
anyone know how to fix??
RuntimeError: Expected tensor for argument #1 'input' ...
is that with my new pretrain?
stuck bane
no
kind ether
what folder is index file supposed to be in
rare gobletBOT
Ayo? @kind ether level 2 !!!
modern oxide
what folder is index file supposed to be in
in logs folder of your model
molten root
did you find a solution? im having the same problem
in my commander, I saw the error message that it couldn't find a certain file. I noticed that it had, strangely enough, a different name from the one I'd entered in "Enter the experiment name", so I just changed the name in "Enter the experiment name" to match the name indicated in the error message.
I had originally put "Zuko test" but it was written in my command terminal that it couldn't find the "Zuko" file, so I changed my "Zuko test" to "Zuko". Since then, it's been working.
brittle wing
hope this is the right chat for it tbh ,
uuhh, no matter what i do in the um. Colab? easy gui thingy i keep getting this error message by the step 2. download model, like it doesnt matter what i put into the url in that i just get this 
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-12-990d181741b5> in <cell line: 4>()
2 #@markdown Link the URL path to the model (Mega, Drive, etc.) and start the code
3
----> 4 from mega import Mega
5 import os
6 import shutil
ModuleNotFoundError: No module named 'mega'
NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.
To view examples of installing some common dependencies, click the
"Open Examples" button below.
proper shale
in my commander, I saw the error message that it couldn't find a certain file. I...
No spaces on experiment names
Basically just that
proper shale
hope this is the right chat for it tbh ,
uuhh, no matter what i do in the um. Co...
Use the newer Colabs:
-colab
proper shale
Use the Ilaria RVC Huggingface space if you intend to infer only, RVC Disconnected or EasyGUI if you wanna train
brittle wing
infer?
proper shale
converting vocals
brittle wing
aha
rare gobletBOT
Ayo? @brittle wing level 1 !!!
proper shale
converting vocals
or anything that involves speaking audio conversion, basically
stuck bane
Use the Ilaria RVC Huggingface space if you intend to infer only, RVC Disconnect...
Why are you recommending that to everyone and not recommending mangio or applio
brittle wing
is there any specifically for singing?
proper shale
Why are you recommending that to everyone and not recommending mangio or applio
Because most people don't have the computers to run them
proper shale
is there any specifically for singing?
U can use any inference colab/space for singing, sooo
stuck bane
That’s being racist to those that do
proper shale
That’s being racist to those that do
...If they're using a Colab, chances are they don't have a good enough PC
brittle wing
andd, which of these do i use?
stuck bane
So you’re body shaming
brittle wing
...If they're using a Colab, chances are they don't have a good enough PC
i sure as hell dont have a good enough pc
molten pecan
andd, which of these do i use?
Ilaria RVC. https://rentry.org/ilarvc_inf_guide here's my guide for it :3
proper shale
andd, which of these do i use?
As I said, Ilaria RVC if you want to infer
proper shale
Ilaria RVC. https://rentry.org/ilarvc_inf_guide here's my guide for it :3
Yeah definitely check out that guide
brittle wing
Ilaria RVC. https://rentry.org/ilarvc_inf_guide here's my guide for it :3
!! will do
proper shale
So you’re body shaming
😭 what
molten pecan
!! will do
Have fun! ^-^
proper shale
😭 what
Look, I always ask them if they have a good enough PC in case they are beginning to use RVC. If they already use it on Colab we can make an assumption
molten pecan
Look, I always ask them if they have a good enough PC in case they are beginning...
He's just trolling
Dw
proper shale
😭
proper shale
Is it a Google Drive link?
brittle wing
yes
dense drift
i created my own model and when i use it it says this
azure marshBOT
-huggingface
dense drift
Upload to Huggingface
i did
proper shale
I was answering Kinero
dense drift
ohh
proper shale
Anyways I'm unsure about the index stuff
young halo
Can someone help me to get rid of the artifacting on my model?
brittle wing
i have no clue what any of this means.
half cove
i created my own model and when i use it it says this
what is the name of ur index
proper shale
i have no clue what any of this means.
O no this is in case you wanna upload to #1175430844685484042
dense drift
what is the name of ur index
half cove
lmfao i was writing how to do it all 😭
half cove
thats why
dense drift
where can i find the added index?
brittle wing
lmfao i was writing how to do it all 😭
the effort is appreciated 😭
dense drift
mz google drive is a mess
proper shale
so i have to be at this part?
Yeye
half cove
mz google drive is a mess
find your rvcdisconnected folder
proper shale
where can i find the added index?
inside the folder with ur model's name
dense drift
thanks
brittle wing
i dont have a normal copy link to tap like it tells me to?
rare gobletBOT
Ayo? @brittle wing level 2 !!!
proven hill
EGIRL AND OTHER GIRL VOICE MODELS HERE!! 
E-Girl, Mommy, V-Tuber, Anime girl and much more!
Wondering what models?
You find them here: https://ko-fi.com/ilariaowo/shop 
INCLUDING FREE ONES!! 
I also do commissions if you want aspecific voice! 
I accept the following payment methods:
- Ko-Fi
- PayPal
- Revolut
- Steam Gift Cards
- Amazon Gift Cards
- Throne
Grab yours today!!!
proper shale
i dont have a normal copy link to tap like it tells me to?
copy link address
spare shale
Yes.. Now I can’t Install the model because it says I need access via link but I’ve already opened it
dense drift
its correct
half cove
or am i doing it entirely wrong
half cove
that did work
gotta copy link address from the download button
half cove
Why don't mine work??
what didnt 
spare shale
what didnt <:nails:1159569314848972891>
I copy the link via google drive but it still says no it doesn't work
half cove
send me the link rq
spare shale
Access is open
half cove
are you using huggingface?
spare shale
send me the link rq
This is the Problem, these are the Voices of My Friends and they Asked Not to Send to Other People Sorry
half cove
oh
spare shale
are you using huggingface?
Google drive
half cove
let me try somthing rq
spare shale
I can show you via Screen Sharing
spare shale
But I won't Talk 😅
half cove
dw
spare shale
patent pasture
aren't there huge differences between Mainline and Mangio RVC?
spare shale
And I understand English very poorly, I only know how to write 😅
patent pasture
And I understand English very poorly, I only know how to write 😅
but it isn't too bad
spare shale
all
half cove
ok then
try using
hmm
go on the zip file to see if its uh formatted correct
oh wait
did it work?
its using trained index
that might be the reason but idk
spare shale
T-T
half cove
let me send you a model and lets see if that works
half cove
same problem is happening to me lol
soft stratus
@proper shale
How i upload voice model in RVC
soft stratus
upload voice?
Yes
soft stratus
I mean how i upload voice model to hugging face
soft stratus
From google drive
spare shale
spare shale
From google drive
This is the Problem Now
soft stratus
I downloaded the model from drive It's not show me my model
spare shale
I downloaded the model from drive It's not show me my model
Yes it is Now and there is a Problem and I don’t understand why
soft stratus
When it'll fix
spare shale
But he's not on the list
spare shale
When it'll fix
sowwy idk
soft stratus
But he's not on the list
That happened with me
spare shale
Sorry, I really don't know
soft stratus
Are you fix this problem
spare shale
I'm not sure 😅
@half cove? 😅
Wait, I'll try to think of something but I'm not sure
@soft stratus
Now it's better to send Models to hf
opal prairie
where can i make models now? all of the colabs aren't working for me
spare shale
@~@
soft stratus
Now it's better to send Models to hf
What it means
soft stratus
How i upload to it
spare shale
How i upload to it
I can show
soft stratus
Show me
soft stratus
Show me again
spare shale
soft stratus
It had with me error in Ilaria
spare shale
??
spare shale
Have you made it private?
soft stratus
How i use it with mobile
soft stratus
Yes
soft stratus
How i copy link with phone
spare shale
dm
Send me a dm and I'll copy the link to you (Sorry, I don't know how to do it via phone)
soft stratus
molten root
hi,
i trained a model locally and it created me more than one hundred .pth, how to know which to use for the inference ?
spare shale
hi,
i trained a model locally and it created me more than one hundred .pth, ho...
The more the better (Good 250 - 500)
azure marshBOT
RVC Guides (How to Make AI Cover)
proper shale
The more the better (Good 250 - 500)
No, just no
proper shale
hi,
i trained a model locally and it created me more than one hundred .pth, ho...
Check the latest epoch ones
- check TensorBoard too
molten root
I don't have tensorboard and I don't want to complicate my task with yet more tools, so I'll take the latest one? ok thanks
proper shale
I don't have tensorboard and I don't want to complicate my task with yet more to...
Alright then
molten root
for the moment I find it quite complicated to use for a not so great result.
proper shale
You're welcome :)
molten root
are there any good quality examples? where you can hear the basic voice (the dataset) and also hear what it sounds like on the final result?
proper shale
Don't think there is a simple one other than just... testing out the model with an audio
molten pecan
How i use it with mobile
molten root
Don't think there is a simple one other than just... testing out the model with ...
and you, have you trained voices and applied this to an audio and are you satisfied with the result? if so, do you think you could share this? i'd really like to know if it's possible to achieve something that really resembles the original voice or not.
proper shale
and you, have you trained voices and applied this to an audio and are you satisf...
well uh yes it is possible, and honestly i dont really want to share 💀
but yes, it's definitely possible
molten root
well uh yes it is possible, and honestly i dont really want to share 💀
oh sorry, I didn't mean to be indiscreet, I apologize, but if you say it's possible, I guess I just have to vary the epoch or inference parameters to improve the rendering.
however, with a 52-minute dataset, quite clean I think, and an epoch of 500, I'm really not satisfied with the result, is that all we can do to adjust our model?
proper shale
oh sorry, I didn't mean to be indiscreet, I apologize, but if you say it's possi...
i mean, yeah, if you dont want to remake your dataset
try with a smaller dataset, maybe
ionic roost
So I've noticed that my main issue seems to be seperation of vocals and instruments
Is there a specific guide I can use for the best way to do this? I use UVR5 right now, should I swap to the RVC default?
half cove
So I've noticed that my main issue seems to be seperation of vocals and instrume...
what model are you using from that rn?
uvr model
half cove
MDX, I normally do this
try mdx32c
ionic roost
I select it with the other two I assume?
ionic roost
Alright
rocky viper
main lintel
I also would like help with making the voice set up i am failing greatly
half cove
depends on ur audio
brittle wing
what are some programs that I can use to make rvc voice models? I tried to use the the tutorial by raw onions but the new version is just a different program entirely
gusty nimbus
anyone else getting the “cannot get file from this private link” on a public google drive link?
rare gobletBOT
Ayo? @gusty nimbus level 1 !!!
rotund prawn
depends on ur audio
6 minutes dataset😶
azure marshBOT
All-In-One Guide on how to make a good model
This guide explains how the D and G files works and much more: https://rentry.org/RVC_making-models
Credits: LUSBERT 
Automated Overtraining Detection (AOD)
Will be available soon in #1159513888199540817
Credits: grvyscale
half cove
look at the sample rate bit
stuck bane
its just staying right here
it's been like this for a few mins
i used cpu to do feature extraction
should i have picked rvmpe_gpu
sand iris
it's most likely not doing anything rn
stuck bane
why
sand iris
unless Applio changed stuff I'm not aware of
sand iris
hmmm
stuck bane
lmao
i think applio installer exe is just broken
even though its the only thing thats letting me boot
rare gobletBOT
Ayo? @stuck bane level 8 !!!
stuck bane
now i just got this
sand iris
brittle wing
wth is happening
idk
brittle wing
wth is happening
im not a tech expert
stuck bane
well i wasn't asking you so
brittle wing
well i wasn't asking you so
Oh
stuck bane
i was asking ppl who would know
stuck bane
<:hmmmm:726807504876601395>
im reinstalling nvidia dependiencies, is this normal?
should i change the code
sand iris
Seems normal
The cuda error is really weird though
I'm not even sure how it could happen
stuck bane
Seems normal
RuntimeError: Cannot install on Python version 3.11.5; only versions >=3.7,<3.11 are supported.: The package you are trying to install has a restriction on supported Python versions. In this case, it cannot be installed on Python 3.11.5. The supported versions are specified as ">=3.7,<3.11," meaning any version of Python from 3.7 up to (but not including) 3.11 is supported.
so i have mini conda (anaconda)
along with python
i only installed it because it needed miniconda.exe in order to run some things
sand iris
python 3.11 will die on you
for RVC related stuff
no point in continuing on that version
Even if you get everything installed, it doesn't run
stuck bane
python 3.11 will die on you
I have python 3.98 though
sand iris
it's running on 3.11 somehow
stuck bane
It’s miniconfa that fucker me over
main lintel
how do i know if my device can run the vc
because i am facing alot of issues and i am doubting my pc abilites
brittle wing
How do I install RVC Local? I download from the github and can't find the option to launch it?
proper shale
Maybe, I'd say keep going for a bit to confirm the suspicion
Or deny it, who knows
Nope, that YT vid is wrong
That was needed back when we used Sovits-SVC
Like a year ago.
ionic roost
Seeing something from 8 months ago being considered old and outdated really just proves how fast technology is progressing
jaunty shale
Question for UVR: Is there any model that removes choir in the background? I already have DeReverb and DeEcho, but I can't find one that removes the choir, or other voices in the bg. Is there any?
proper shale
Question for UVR: Is there any model that removes choir in the background? I alr...
Karaokee models probably
jaunty shale
Karaokee models probably
could I get exact name of the model? if it's VR or MDX
I am not certain in which category and which one it is
proper shale
could I get exact name of the model? if it's VR or MDX
There's MDX karaoke, and it's also Karaokee in VR
proper shale
There's MDX karaoke, and it's also Karaokee in VR
But then again I forgot what it's called cuz i just use MVSep
main lintel
how do i know if my device can run the vc
because i am facing alot of issues and...
am once again asking what i should do to check if my pc can handle it
proper shale
how do i know if my device can run the vc
because i am facing alot of issues and...
what's your gpu
proper shale
Quad pro K2000
Oh that's a Quadro card??? Hmm let me check wait
main lintel
yea i have it for graphic things not gaming things
rare gobletBOT
Ayo? @main lintel level 1 !!!
main lintel
Ayo? <@604507630047592478> level **1** !!! <:lfg:1159355870119993496>
Ayo!?
proper shale
Yeah no that ain't gonna cut it
main lintel
damn
proper shale
Way too old yeah
jaunty shale
There's MDX karaoke, and it's also Karaokee in VR
for MDX i found karaoke 1 and 2, same in VR, which is 5_HP-Karaoke-UVR and 6_HP-Karaoke-UVR Do you know which one is better?
main lintel
Way too old yeah
wait what about K2200?
proper shale
for MDX i found karaoke 1 and 2, same in VR, which is 5_HP-Karaoke-UVR and 6_HP-...
Karaoke 2 on MDX might be a good one
About VR, I'm not sure, I suggest trying out both
proper shale
it is old?
Yeah - 10 years old at this point
jaunty shale
Karaoke 2 on MDX might be a good one
About VR, I'm not sure, I suggest trying o...
okay, I'll try and see if it works
molten pecan
for MDX i found karaoke 1 and 2, same in VR, which is 5_HP-Karaoke-UVR and 6_HP-...
5 and 6_HP work the same, except 6 has a larger extraction depth
main lintel
Yeah - 10 years old at this point
ouch am getting old
proper shale
K2200 would also not be enough... 4GB VRAM and older architecture might be a no no
main lintel
it worked with me though
proper shale
Minimum specs would be an equivalent to a 1060, ig
main lintel
maybe there's another thing i could do for the K2000 to work?
the K2200 worked for me
at work that is
proper shale
it worked with me though
Might be because of the extra CUDA cores in those
stuck bane
5 and 6_HP work the same, except 6 has a larger extraction depth
They work the same except you use 6hp after deverbing and deecho
main lintel
am not sure what is Cuda
jaunty shale
They work the same except you use 6hp after deverbing and deecho
okay, ill try that as well, thank you
proper shale
Might be because of the extra CUDA cores in those
...well the K2200 has half the CUDA cores of the GTX 1060
molten pecan
They work the same except you use 6hp after deverbing and deecho
yeah of course
proper shale
(CUDA cores are kinda needed for this AI workload)
stuck bane
(CUDA cores are kinda needed for this AI workload)
What’s the cuda cores in a 2060 than a 3070
(Asking for my dog)
proper shale
What’s the cuda cores in a 2060 than a 3070
2060 has like ~2000, 3070 has 5888
stuck bane
Rtx right
stuck bane
What abt a 3070 and a 3080
proper shale
3080 has 8k
stuck bane
4060
proper shale
3072
stuck bane
Decent
proper shale
Better than a 2060 definitely
proper shale
I think for a 3080, 8 is perfect
3070 would have to be 6 i guess
Same for 4060
and for 2060 go for 4
This is just considering VRAM limitations, I believe those might not be the best for your model, as every model has its perfect batch size
stuck bane
True
sand iris
maybe there's another thing i could do for the K2000 to work?
the K2200 worked f...
k2000 or k2200 ?
stuck bane
If I can get applio to work then I’ll use ur advice
sand iris
k2000 won't work. period.
stuck bane
Thanks!
stuck bane
k2000 won't work. period.
Is there amd support yet
sand iris
Is there amd support yet
For what ?
stuck bane
Applio
stuck bane
You're welcome :)
More batch size faster the epoch right
proper shale
More batch size faster the epoch right
Yes
torn wasp
I already feel that this does not work, I do the steps correctly and I end up getting an error.
proper shale
I already feel that this does not work, I do the steps correctly and I end up ge...
Try Ilaria RVC, I've heard some Huggingface Spaces wouldn't work and weren't fixed
torn wasp
perform the steps in order, and if it comes out wrong, sometimes when loading the voice it comes out with a male voice being Michael Jacksons
yes, now I get errors from this, so which link or google colab is more recommendable and which gives pretty perfect results?
stuck bane
yes, now I get errors from this, so which link or google colab is more recommend...
If you’re trying to infer only
Replay will be your best bet
proper shale
yes, now I get errors from this, so which link or google colab is more recommend...
No such thing as a "perfect" result
torn wasp
I have done a michael jackson AI before with a google colab link, which I thought was perfect.
proper shale
I have done a michael jackson AI before with a google colab link, which I though...
Oh I've done countless covers with his vocals
torn wasp
I could hear Michael Jackson's voice quite naturally without having to listen to a robotic voice, here is the video that I am still saving.
stuck bane
I have done a michael jackson AI before with a google colab link, which I though...
Well, the most perfect result with your model, will be the indexing to the max, and if you just want to convert without any errors, download replay will serve your request
proper shale
I could hear Michael Jackson's voice quite naturally without having to listen to...
chances are, you're not increasing the pitch as much as the song needs
stuck bane
✅
torn wasp
but when using this interface, which I showed you, the first thing is that being at 0, the voice seems very masculine as if it were an effect, I raise it to 2 and an error appears, and it really does not work.
proper shale
Had to delete due to copyright
proper shale
but when using this interface, which I showed you, the first thing is that being...
Again, try using Ilaria RVC
torn wasp
Had to delete due to copyright
Uh
proper shale
-colab
torn wasp
Well, of all of these, which one did you find most effective?
Are all those called "Google Colab" all the same when performing the AI, or are there google colabs that work more on the AI to match the voice I insert
proper shale
im back from earlier today, what does these things mean exactly?
The options on top are the pitch extractors, RVMPE is the best really
Median filtering does not need to be used as it doesn't work in any other pitch extractor BUT harvest
Second one is if you want the volume to be closer to the dataset levels or to be as close as the audio you're converting.
Third one protects voiceless consonants, never really found it that useful
torn wasp
AICoverGen-WebUI, modded by Hina Google Colab
AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
AICoverGen-NoWebUI [Spanish], credits to Eddy, Hina and Gdr for translating and fixing Google Colab
RVC Disconnected, by Kit Lemonfoot Google Colab
easyGUI, by rejects Google Colab
proper shale
Well, of all of these, which one did you find most effective?
...Ilaria RVC? I literally just told you to use it
torn wasp
of these, which is the best
proper shale
Are all those called "Google Colab" all the same when performing the AI, or are ...
They're all the same under the engine
proper shale
of these, which is the best
There is not really a "best" one, it all depends on your use cases
torn wasp
...Ilaria RVC? I literally just told you to use it
Ok, please pass me the link, Thank you
proper shale
... it's right there
proper shale
In case you have any doubts
And actually I strongly recommend reading this while you set it up
torn wasp
https://www.rentry.org/ilarvc_inf_guide
oh, ready, it's the tutorial you mentioned to me.
and a doubt, if you ask me about those numbers from 0 to so and so, which one fits well when we talk about Michael Jackson.
proper shale
and a doubt, if you ask me about those numbers from 0 to so and so, which one fi...
Depends on the vocals
proper shale
rvmpe is fast and prolly best
main lintel
...well the K2200 has half the CUDA cores of the GTX 1060
ah understandable
main lintel
k2000 or k2200 ?
i have both
k2000 is the one that is not working
also sorry for the late reply the internet cut off
sand iris
i have both
k2000 is the one that is not working
Yeah unfortunately you won't get it working as it's simply too old
Also don't stress over not replying 😛 all good
sand iris
From what I remember the gtx 1xxx cards give poor results
Even though it will work, don't expect low latencies
You could give google colab a chance
main lintel
From what I remember the gtx 1xxx cards give poor results
i mean the Quad pro K2000 works
rare gobletBOT
Ayo? @main lintel level 2 !!!
clever geode
is there any way to extract 1 person's voice from a video in which multiple people speaking one by one, for eg and interview
like using some Ai
sand iris
it just has abit of lag
The 1050 will barely make latency lower
sand iris
Check out the colab notebook in the pins at #🔍│help-w-okada , maybe it could work well enough
main lintel
can you explain me what collabs are?
main lintel
is there any way to extract 1 person's voice from a video in which multiple peop...
if it will help if they are speaking seprtaly you could cut the video on an editor
sand iris
can you explain me what collabs are?
Colab is a google platform to run stuff related to machine learning
People made voice-changer run on them, so you can (ab)use one of google's GPUs to run it
So no meaningful load on your machine as everything is done remotely
sand iris
Unsure about the values, people use 6 on 8GB cards
I think you can do up to 12
If it crashes, lower it by like two
clever geode
i usually do it on 14
i wonder if i can do more
but i dont wanna be crashed either
half cove
clever geode
so i thought if anyone knew maximum
torn wasp
I don't really understand, I'm using this link to turn a voice into Michael Jackson
Do it again by setting it to "2" and the voice sounds like the original voice of the song
How do I do in this case?
torn wasp
Depends on the vocals
.
torn wasp
-colab
azure marshBOT
RVC Guides (How to Make AI Cover)
spiral osprey
I can't complete my voice rendering with rvc disconnected, I re-rendered up to 1000 epouchs, it didn't reach 500 and stopped, as far as I know, there was no gpu limit error because it had already been more than 24 hours before rendering the voice, I don't know what This is happening, this is the 3rd time this has happened. Can anyone help me the 2nd time and the 1st time was gpu limit. Now I don't know but I don't have a print to show. Can someone help me?
proper shale
I can't complete my voice rendering with rvc disconnected, I re-rendered up to 1...
You probably weren't actively checking the page
spiral osprey
I was on the page a while ago. and the afk message did not appear.
glad zealot
I can't complete my voice rendering with rvc disconnected, I re-rendered up to 1...
- Probably run out of gpu already since you said the first time it gave the warning
- Not checking the page once in a while(afk)
spiral osprey
I was on the page a while ago. and the afk message did not appear.
he was no longer counting the running time.
1- the page was rendering normally.
2- an afk message did not appear.
proper shale
You should limit your training sessions to an hour or so
To prevent losing your model like that.
glad zealot
Colab without pro can only go around roughly 4 hours(less most of the time)
spiral osprey
You should limit your training sessions to an hour or so
what do you mean limited? 1 hour and I had already waited more than 24 hours due to the GPU.
glad zealot
After that it will run without gpu
proper shale
what do you mean limited? 1 hour and I had already waited more than 24 hours due...
I mean, stop training after an hour or so of training and export the model so you don't lose it.
spiral osprey
how to export the model. Or is there a way to continue where I left off?
proper shale
how to export the model. Or is there a way to continue where I left off?
There's an Export Model step
And yes, you can continue from where you left off, if you use the Export model step
spiral osprey
and how do I export it and continue where I left off?
proper shale
and how do I export it and continue where I left off?
To export, just run the step
Then once you come back, you set all of the same Training Variables, then use only the steps marked (for resuming) and then just train model with the same batch_size you had previously
spiral osprey
Do you have any guide or printout? and to be able to continue where I left off.
proper shale
Do you have any guide or printout? and to be able to continue where I left off.
Well, there isn't really a guide for resuming, but what I wrote should be enough.
It really isn't hard, just gotta get used to it.
spiral osprey
Well, there isn't really a guide for resuming, but what I wrote should be enough...
ok thanks,and that I don't usually use the website much.
proper shale
ok thanks,and that I don't usually use the website much.
You're welcome ;)
Let me know if you have any hassles and I'll guide you through em
spiral osprey
look private chat.
rare gobletBOT
Ayo? @spiral osprey level 7 !!!
glad zealot
I mean, stop training after an hour or so of training and export the model so yo...
Might need your help to add that feature to kaggle
proper shale
look private chat.
I'm sorry but I don't really like helping in DMs...
glad zealot
Make it work OwO
acoustic light
does anyone know how to use this? https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb
spiral osprey
I'm sorry but I don't really like helping in DMs...
oh,ok so sorry
lone abyss
yalll question, is there a way to add a custom pretrain that has a higher range than usual? let's say 6th octave. If yes how do I do it? thanks in advance xx
proven hill
EGIRL AND OTHER GIRL VOICE MODELS HERE!!
E-Girl, Mommy, V-Tuber, Anime girl and much more!
Wondering what models?
You find them here: https://ko-fi.com/ilariaowo/shop
INCLUDING FREE ONES!!
I also do commissions if you want aspecific voice!
I accept the following payment methods:
- Ko-Fi
- PayPal
- Revolut
- Steam Gift Cards
- Amazon Gift Cards
- Throne
Grab yours today!!!
violet heron
Are you using RVMPE?
rare gobletBOT
Ayo? @timber solar level 2 !!!
violet heron
Models converting stuff in other languages can be a bit iffy
You can try turning down the index by a bit
You can’t share here
lime sorrel
if i want to make a model, can i put a full 1-2 minute audio inside the RVC or do i have to cut each audio part?
rare gobletBOT
Ayo? @lime sorrel level 1 !!!
tame mica
real
glad zealot
if i want to make a model, can i put a full 1-2 minute audio inside the RVC or d...
You can put that but make sure it's clean, no background noise or other voices
heavy shuttle
if i want to make a model, can i put a full 1-2 minute audio inside the RVC or d...
got to be in segments of like 10 second clips, people say that the ideal amount of time to get a decent model is 10-20 minutes of data
but with this new pretrain they say its pretty cracked...
violet heron
got to be in segments of like 10 second clips, people say that the ideal amount ...
No they don’t
RVC splits them automatically and you don’t need to do that
tame mica
RVC splits them automatically and you don’t need to do that
dont*
heavy shuttle
No they don’t
well the clips cant be too long, i remeber doing them in one minute segments and it gave me a length error.
tame mica
violet heron
well the clips cant be too long, i remeber doing them in one minute segments and...
Skill issue then because I’m able to train a single 12 minute audio file
heavy shuttle
Skill issue then because I’m able to train a single 12 minute audio file
and it works??? you dont even need to cut them? what on earth how?
rare gobletBOT
Ayo? @heavy shuttle level 3 !!!
violet heron
and it works??? you dont even need to cut them? what on earth how?
You don’t
Back in the SVC days yes
rare gobletBOT
Ayo? @violet heron level 90 !!!
heavy shuttle
You don’t
I thought that the only reason why you couldn't do that was it wouldnt handle it well , would make it sound weird
violet heron
I thought that the only reason why you couldn't do that was it wouldnt handle it...
Nope, it wouldn’t at all
The reason it might because RVC doesn’t fair well to long pauses/silences
heavy shuttle
The reason it might because RVC doesn’t fair well to long pauses/silences
Ohhhhh ok makes sense.
apprecaite this info
now i dont gotta cut stuff lol 😂
violet heron
now i dont gotta cut stuff lol 😂
You do have to trim silences over 1 second , just not separate files
lean owl
i downloaded a model i made and tried refreshing and this piece of shit still doesnt show up
violet heron
i downloaded a model i made and tried refreshing and this piece of shit still do...
Did you put your .pth file in the weights folder
And the .index in logs
lean owl
its a huggingface link
violet heron
its a huggingface link
Did you duplicate the space
lean owl
yes i did
violet heron
And did you put any spaces in the name
lean owl
well i just titled it this
violet heron
well i just titled it this
No I’m saying the model name
lime sorrel
got to be in segments of like 10 second clips, people say that the ideal amount ...
I've heard that RVC v2 automaticly cuts the audio
lean owl
No I’m saying the model name
oh no
lime sorrel
i dont want to take more hours to cut every single individual soundbytes
btw how much epochs do i set in if the total audio lenght is around 5 minutes long?
glad zealot
you can use audacity to do it automatically
lime sorrel
you can use audacity to do it automatically
?
glad zealot
to cut the audio
lime sorrel
how?
glad zealot
also the trainer will automatically do that
lime sorrel
does audacity really let you do that?
violet heron
no i didnt
What was the link
lime sorrel
use the label feature
ok, ill try
fluid topaz
btw how much epochs do i set in if the total audio lenght is around 5 minutes lo...
i would try 300, then look at the graph
glad zealot
violet heron
What was the link
If it has “blob” in it change it to “resolve”
lean owl
What was the link
a google drive link. idk if thats allowed for huggingface spaces
violet heron
a google drive link. idk if thats allowed for huggingface spaces
Not anymore
Has to be a HuggingFace one
lean owl
oh course it does omfg
lean owl
Not anymore
oh yeah which method do i use
glad zealot
ne wmodel
lime sorrel
this is my result
i think the sounds are to close
should i modify the minimum and maximum?
lean owl
i love audacitys confusing menu
lime sorrel
editing it it to 10 milisec fixed it
lime sorrel
so exporting the labels, it wants to export it as a txt file, is this the right thing to do?
rare gobletBOT
Ayo? @lime sorrel level 2 !!!
lime sorrel
this?
worldly adder
Is it possible to put mp3 audio instead of using the microphone?
glad zealot
yes
violet heron
Is it possible to put mp3 audio instead of using the microphone?
Yes
glad zealot
inference
violet heron
Is it possible to put mp3 audio instead of using the microphone?
wav better*
glad zealot
mp3 is fine
worldly adder
but I'm talking about not having the option to add (sending the photo)
rare gobletBOT
Ayo? @worldly adder level 1 !!!
blissful timber
RuntimeError: Expected tensor for argument #1 'input' to have the same type as tensor for argument #2 'weight'; but type torch.cuda.FloatTensor does not equal torch.cuda.HalfTensor (while checking arguments for cudnn_batch_norm)
having this problem when using the rmvpe_gpu extraction
scarlet relic
so i decided to try mangio for real time conversion, however i dont hear anything at all when i press start audio conversion. However the "inference time" does fluctuate which probably means that it is converting i just dont know why i cant hear it
raven osprey
Am i supposed to convert the zip files to google drives and megas myself?
I'm stuck on step 2
for the google collab rvc
violet heron
Am i supposed to convert the zip files to google drives and megas myself?
Yeah if it’s a weights link
HuggingFace link no
raven osprey
Im tryna do perfect cell from dbz
i downloaded them and tried converting
but still stuck
https://huggingface.co/EpicStar86/EpicStar86RVC/resolve/main/NewTFSPerfectCell.rar?download=true so this link right here, how would I input it in step 2?
it says the mega error
violet heron
https://huggingface.co/EpicStar86/EpicStar86RVC/resolve/main/NewTFSPerfectCell.r...
it’s a rar
raven osprey
ye
violet heron
ye
Download it unzip the contents, rezip them as a .zip
Upload to drive/mega/HuggingFace and copy the link
Make sure it’s public
raven osprey
ive done the drive one but it still says its not public when i put it to allow link stuff
i can send it if you want
rigid light
so is just rmvpe_gpu broken in applio
violet heron
ive done the drive one but it still says its not public when i put it to allow l...
Make it public
warped hedge
is this the correct video for ai voice changer? https://www.youtube.com/watch?v=_JXbvSTGPoo&ab_channel=AIToolsSearch
violet heron
is this the correct video for ai voice changer? https://www.youtube.com/watch?v=...
I wouldn’t really watch videos
They often skip stuff
But yes it is
warped hedge
oh i see, is there anything in here i can use to know how to do the steps correctly?
tame mica
-rvc
tame mica
^