internet issue tbh
#✨│ai-help
merry spindle
ornate ice
once the internet is on, maybe try finding a local training that does not do the internet
ornate ice
internet issue tbh
yeah it's probably lag
zinc anchor
The user who made the RVC Disconnected notebook said it's a Chrome issue
merry spindle
ah
merry spindle
The user who made the RVC Disconnected notebook said it's a Chrome issue
do i have to restart
like is it a HUGE issue
zinc anchor
Not huge for me, you can still train, you just won't be able to see the metrics
merry spindle
i dont give a singular shit about the metrics
is 500 epochs alot because it made me train 500 epochs as the default
ornate ice
500 epoches is kind of a lot, but it's probably necessary
merry spindle
is there a way to delete all things about the training so i can redo the training entirely..?
small zodiac
can anyone dm and help me on how to use RVC v2 Hugging Face Edition?
merry spindle
what.
zinc anchor
Maybe openrail
rare gobletBOT
@merry spindle has reached level 4.
merry spindle
you are a helper you should know this
zinc anchor
@eager pelican
merry spindle
you have to upload them to huggingface right
zinc anchor
Yeah
merry spindle
please
zinc anchor
Use openrail
merry spindle
okay
serene hound
i think it was better than before, it's now less "loss"... i think it may work now?
eager pelican
<@917711764571951144>
wha
merry spindle
am i supposed to upload a folder with both the index and the pth file?
eager pelican
zip them
merry spindle
okay
does it have to be only the pth and the index
can i have the rvc for making covers
i ean colab
zinc anchor
wha
The openrail stuff
Do the people who say it's bad have a concrete reason to say that?
somber oracle
so for some reason my colab console said that a pth filed was successfully copied to my drive
but the file was nowhere to be found in my actual drive
eager pelican
Do the people who say it's bad have a concrete reason to say that?
nope they just say its bad and never say why
lemme look into ig
dont have sentences instantly cut off
basically have it fade away
eager pelican
Do the people who say it's bad have a concrete reason to say that?
bing explains it better either way
but yea ig its simply not perfect
but there seemingly is no better alternatives either
ah yes no embed but well its a wall of text either way
rare gobletBOT
@rigid sky has reached level 1.
wispy burrow
eager pelican
oh realtime
well thats just the settings
For realtime go to #🔍│help-w-okada
oblique venture
can someone explain me what all these configs for
and what is the diffrence of each pitch extraction algorithm
glossy coyote
hi did you manage to fix this? I'm having the same issue
I can't seem to get that runtime folder
amber jolt
Hello, good day, so I've been training a voice model and while the quality of it is fine, the voice is way different from the original. please help. 
also, is there a way to fuse different voices?
junior shell
I'm getting so mad that all these methods just refuse to work. Just yesterday, I managed to get this method to work, I managed to finally make a new voice model. So I tried to make another one, it was working, but it was taking longer than the first and I had to shut off my computer for the night. So I had to restart the whole, all the steps went by fine, but 4.Train Model, it just can't seem to work as it did before. I don't know if interrupting it had anything to do with it. It gave the Index, but it's not making the pth
https://drive.google.com/file/d/1uSJ0qCnh0-Fo28Zk6DKbSFsNWdLJXURm/view?usp=sharing
brittle wing
anyone know what is this TTS? there's a Emotion voice feature at it
brittle wing
hi did you manage to fix this? I'm having the same issue
Nope. I have no idea how to fix it
glossy coyote
Nope. I have no idea how to fix it
: (
brittle wing
lemme look into ig
You my bro Lusbert
Please help :<
Damn, he is not responding. he is dead :<
rare gobletBOT
@brittle wing has reached level 3.
eager pelican
You my bro Lusbert
i was making dataset
amber jolt
Trying to research answer to my own questions in the meantime jfkldjfd
brittle wing
i was making dataset
My Applio Installer is not working anymore. It keep giving me a error
This error
eager pelican
My Applio Installer is not working anymore. It keep giving me a error
applio is pretty stupid
just dont use it and if you want to well idk how to fix it
it never works and never will
amber jolt
what is applio even for?
rare gobletBOT
@amber jolt has reached level 1.
amber jolt
like what's the diff
brittle wing
applio is pretty stupid
I know, but Applio is the only one that have AMD support and crepe. So I'm stuck using Applio
It works before and now it's broken for no reason
Ugh
eager pelican
¯_(ツ)_/¯
amber jolt
Hello, good day, so I've been training a voice model and while the quality of it...
Don't suppose you'd know the answer to this problem? ;w;
amber jolt
anybody still here to help?
brittle wing
anyone know what is this TTS? there's a Emotion voice feature at it
why do my questions always get skipped? 😑
amber jolt
I'm feeling that too
sand iris
I know, but Applio is the only one that have AMD support and crepe. So I'm stuck...
no
The original RVC also has the same amd support
glossy coyote
The original RVC also has the same amd support
;o
rare gobletBOT
@glossy coyote has reached level 1.
brittle wing
The original RVC also has the same amd support
Really? With crepe?
sand iris
Yeah they have crepe too
amber jolt
Might I trouble you with a question myself @sand iris?
lean vine
I haven't used the program since easy gui was banned from collab which one is currently the best
rare gobletBOT
@lean vine has reached level 2.
brittle wing
Yeah they have crepe too
So where's the link?
glossy coyote
Really? With crepe?
btw i think i've managed to fix it
brittle wing
Applio?
sand iris
So where's the link?
So there's this
https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/blob/main/docs/en/README.en.md
sand iris
But then it doesn't contain great instructions on getting it running
glossy coyote
@brittle wing open the manager and download the runtime from there instead
sand iris
But they have prebuilt archives on their hugging face page
https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main
Relevant file is the latest nvidia .7z archive
brittle wing
But then it doesn't contain great instructions on getting it running
Great puff
sand iris
Said archive should just be unzipped, and it'll just work
glossy coyote
<@456226577798135808> open the manager and download the runtime from there inste...
@brittle wing lemme know if it works for u
brittle wing
<@456226577798135808> lemme know if it works for u
Ah okay im going to take a look that when I get home
Thank you
glossy coyote
np
sand iris
Applio more like trainwreck
brittle wing
~~Applio more like trainwreck~~
Pufff
amber jolt
What's the difference between applio and rvc anyway?
I've tried looking it up but I guess my google fu sucks
sand iris
What's the difference between applio and rvc anyway?
Applio is a derivative of a derivative project (the original project being RVC-WebUI)
Which was supposed to make stuff more user friendly, but instead brought even more issues!!!
amber jolt
Yeah, that's defnintely the impression I'm getting
rare gobletBOT
@amber jolt has reached level 2.
glossy coyote
-rvc
azure marshBOT
RVC Guides (How to Make AI Cover)
sand iris
Gets worse when you consider :
- Applio has multiple installers
- Said installers do wacky shit
- They bug out
- The Applio code itself does wacky shit
- The docs apparently have issues (I've seen one thing where it makes users download several extra useless gigabytes)
All of this for a few convenience features and "user friendliness"
amber jolt
Eeks 
sand iris
Kind of mad we have to deal with a whole lot of different python installs for no reason
amber jolt
So like in terms of output for voice models it's really no different than rvc?
sand iris
So like in terms of output for voice models it's really no different than rvc?
I don't remember the exact details, I think there is one exception, the OG RVC project uses a japanese model to regenerate audio whereas Applio has english available which could lead to better accuracy
i didn't check it, this might be completely false
I just vaguely remember something along those lines
wispy burrow
also, is there a way to fuse different voices?
There is.
CKPT Processing tab in Mangio.
Applio has it too.
I think it's in the Extras tab there.
amber jolt
Aahhhh Sweet. Just gotta figure out what files to put where now cuz I'm dumb
well, file paths
Bear with me here, what do I put for 'model information to be place'? And 'saved model name'?
also Original question: quality of voice model I'm training is fine but the voice is different than the original? Any clue what I"m doing wrong?
wispy burrow
Bear with me here, what do I put for 'model information to be place'? And 'saved...
Model information is something I'd like to know as well.
Saved model name should be the new name for the fused model.
amber jolt
Okay.
wispy burrow
also Original question: quality of voice model I'm training is fine but the voic...
Different as in too high or too low?
amber jolt
This is after I've adjusted pitch to account for female to male, but yeah It's a bit nasally and high and the tone is just off
sand iris
During the first few steps of training, did it detect your audio files ?
wispy burrow
But they have prebuilt archives on their hugging face page
https://huggingface....
RVC0813Nvidia? I can't get it to work.
I get an error...
I'd like to try that one out. I heard they updated their RVMPE model.
sand iris
Mind sharing the error ?
amber jolt
During the first few steps of training, did it detect your audio files ?
this question for me?
sand iris
this question for me?
Yeah
amber jolt
I believe so? is there a way to check if it didn't?
like during feature extraction it does mention the list of wav files from the datasets
sand iris
Oh so it did that step just fine
amber jolt
hold on, I'm looking at something that says .wav contains nan?
wispy burrow
Mind sharing the error ?
C:\Users\User\Documents\AI\RVC0813Nvidia>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
Traceback (most recent call last):
File "C:\Users\User\Documents\AI\RVC0813Nvidia\infer-web.py", line 23, in <module>
import gradio as gr
File "C:\Users\User\Documents\AI\RVC0813Nvidia\runtime\lib\site-packages\gradio_init_.py", line 3, in <module>
import gradio.components as components
File "C:\Users\User\Documents\AI\RVC0813Nvidia\runtime\lib\site-packages\gradio\components.py", line 34, in <module>
from gradio import media_data, processing_utils, utils
File "C:\Users\User\Documents\AI\RVC0813Nvidia\runtime\lib\site-packages\gradio\processing_utils.py", line 22, in <module>
from gradio import encryptor, utils
File "C:\Users\User\Documents\AI\RVC0813Nvidia\runtime\lib\site-packages\gradio\utils.py", line 42, in <module>
import matplotlib.pyplot as plt
File "C:\Users\User\AppData\Roaming\Python\Python39\site-packages\matplotlib\pyplot.py", line 55, in <module>
from matplotlib import rcsetup, style
File "C:\Users\User\AppData\Roaming\Python\Python39\site-packages\matplotlib\style_init_.py", line 1, in <module>
from .core import available, context, library, reload_library, use
File "C:\Users\User\AppData\Roaming\Python\Python39\site-packages\matplotlib\style\core.py", line 26, in <module>
import importlib_resources
ModuleNotFoundError: No module named 'importlib_resources'
C:\Users\User\Documents\AI\RVC0813Nvidia>pause
glossy coyote
does anyone know what this error means?
amber jolt
fffff brb food but I'll have my phone on me
rare gobletBOT
@amber jolt has reached level 3.
sand iris
C:\Users\User\Documents\AI\RVC0813Nvidia>runtime\python.exe infer-web.py --pycmd...
I'll check this later, but it seems like one of the random runtime errors
That happen for no reason at all
glossy coyote
did you manage to find a solution for this?
shy monolith
did you manage to find a solution for this?
its something to do with gradio
never fixed it
i just used okada for realtime
glossy coyote
:/
wispy burrow
hold on, I'm looking at something that says .wav contains nan?
Could be that there is to much silence in the audio file?
wispy burrow
I'll check this later, but it seems like one of the random runtime errors
Appreciate it.
amber jolt
Could be that there is to much silence in the audio file?
Is that what that means?
Cut the silences tho when compiling the stuff
glossy coyote
have you found a solution for this?
rare gobletBOT
@glossy coyote has reached level 2.
sand iris
have you found a solution for this?
Mind sharing your own log ?
I think the issue is generic and you can't really apply other's fixes
wispy burrow
Is that what that means?
Nan stands for "not a number".
To me that sounds like "there's no sound to extract pitch from in certain parts" or something.
I could be wrong though...
amber jolt
Np I’ll take another look at the files
To be clear there can’t be any bits of silence whatsoever?
glossy coyote
Mind sharing your own log ?
sand iris
Which GPU is this running?
sand iris
I could be wrong though...
Honestly sounds like it, I think it was one of the features mangio added, cutting out silence
Applio maybe isn't 200% useless after all
glossy coyote
Which GPU is this running?
I chose AMD
wispy burrow
To be clear there can’t be any bits of silence whatsoever?
Very short pauses are fine. No longer than 0.5 seconds.
glossy coyote
You can't train models on AMD at the moment unfortunately
this is voice cloning tho
amber jolt
Honestly sounds like it, I think it was one of the features mangio added, cuttin...
Yeah, if there are any, it’d be those.
Frick wrong person sry
sand iris
this is voice cloning tho
Oh my bad, sorry
Only seen toBytes happen for training, I assumed this was the case here
Which card is it exactly?
amber jolt
Very short pauses are fine. No longer than 0.5 seconds.
Yeah, I’m looking and there’s nothing that would be more like .5 if any
sand iris
It's complaining about not having enough memory, that's unless because the card just doesn't have enough, or that it's using another GPU in the system (integrated graphics most likely)
glossy coyote
Oh my bad, sorry
and also, is it normal for it to say no nvdia gpu found even tho I chose amd?
sand iris
Yeah that's expected
glossy coyote
It's complaining about not having enough memory, that's unless because the card ...
i'm assuming that it's this? ;o
magic hare
is there some kind of website or smth were I can train voice models easily, yknow savingme from all the tecnical stuff?
sand iris
i'm assuming that it's this? ;o
Oh yeah that's the problem lol
Integrated graphics won't cut it unless you are able to give them a lot of video memory
wispy burrow
Yeah, I’m looking and there’s nothing that would be more like .5 if any
Do you have some very short clips in the dataset? That are less than a second for example?
amber jolt
Do you have some very short clips in the dataset? That are less than a second fo...
Any of those were combined to other clips
wispy burrow
Hmmm...
wispy burrow
Any of those were combined to other clips
Do you still have the logs for preprocess & feature extraction?
wispy burrow
Do they mention some error or something?
amber jolt
None that I'm seeing. I'm checking an earlier attempt that had some diff settings on Mangio-crepe but same problem. Doesn't have the nan in it's feature extraction so I'm guessing that's not the issue after all?
Would it help give you a better idea if I sent like a clip with original voice and generated voice?
wispy burrow
None that I'm seeing. I'm checking an earlier attempt that had some diff setting...
What sample rate did you use? 40 or 48?
Most people seemed to use 40. I read there was a bug with 48.
I read that on the old Hub.
No idea if it got fixed or what the bug actually was...
amber jolt
I did on 40
There are a fair amount of bits in datasets where there is like battle yell/getting hurt sounds or like an injured tone in some clips, could that be it?
Mostly it's just talking but yeah
actually 90% is just talking
wispy burrow
There are a fair amount of bits in datasets where there is like battle yell/gett...
That should be fine.
amber jolt
I am sadness and frustration jfldsj
wispy burrow
I can only think of one more thing...
Spaces or special characters in the path to the RVC folder or in the dataset folder.
wispy burrow
_ and - are completely fine.
amber jolt
Hm, that's the only unique characters
wispy burrow
And no spaces?
amber jolt
no spaces
I used the underscore in place of spaces
wait
hold on
one of the dataset files does have a space
Gonna run a 30 epoch and see what I get
rare gobletBOT
@amber jolt has reached level 4.
amber jolt
honestly thought spaces and stuff would just not make it work but if it actually affects quality and I been doing this the whole time jfkdls
wispy burrow
Any of those were combined to other clips
What software did you use for this? Audacity?
I need to combine some really short files this time.
But never used Audacity. Just to remove silence
amber jolt
I actually used Wavepad Sound Editor but I also have Audacity in case there's something I can only seem to do on there
wispy burrow
Well, I'm slowly figuring it out...
rare gobletBOT
@wispy burrow has reached level 16.
amber jolt
It was a bit overwhelming a first but a bit of youtube videos to find what I wanted and it wasn't really an issue.
OH
did you make sure to get the non-commercial version?
otherwise it'll try forcing you to buy it after the trial. and if not it'll limit you to just being able to save as like a wavepad project file thing
brittle wing
<@456226577798135808> lemme know if it works for u
Oh carp bro! I got it to work!
Thank you
meager edge
anyone on?
I've finally gotten this thing to run locally and I'm reaching this issue no matter what steps I take. I've done this so many times through colab but obviously that isn't possible anymore. I'm not sure what this "VC" is, but I've triple checked everything I've done and my PTH and Index files are where they should be. Not the highest end GPU but I've read you don't need it for conversions, just training.
glossy coyote
Oh carp bro! I got it to work!
YAYYYY
brittle wing
YAAYYYYY
meager edge
If not, I'm getting this error and I'm unsure where to go from here. I've looked these up online to no avail. I saw one thing on github but the responses din't really corelate to my issue directly.
Didn't*
meager edge
Don't mean to bother, where should I add the path? I'm not really seeing a box that lets me add a specific path to any of my files
glossy coyote
If not, I'm getting this error and I'm unsure where to go from here. I've looked...
im having the same error too
are u using amd gpu as well
meager edge
let me check brb
rare gobletBOT
@meager edge has reached level 3.
sand iris
does mangio even support intel/amd ?
wispy burrow
It doesn't.
wispy burrow
I've finally gotten this thing to run locally and I'm reaching this issue no mat...
You have spaces in your path to the RVC folder.
meager edge
I mean- if it doesn't support it, I'm wasting my time. I'll do some more research. I appreciate the time. There's no box on my screen that references that.
timid falcon
how do i resume training on rvc disconnected? is having just a for example just the e220_whatever.pth enough or do i need the logs folder as well?
wispy burrow
I mean- if it doesn't support it, I'm wasting my time. I'll do some more researc...
Inference works on CPU though.
sand iris
I mean- if it doesn't support it, I'm wasting my time. I'll do some more researc...
There's applio and RVC-WebUI that support it though
wispy burrow
I mean- if it doesn't support it, I'm wasting my time. I'll do some more researc...
If you don't have a compatible GPU, Mangio then uses your CPU.
sand iris
Assuming you get Rvc webui working, seemed not to work and I can't check
meager edge
Just curious because I'm new to this- when you say interface, you mean the localhost UI? If the UI itself supports it but not the method of audio conversion, it should still not work. At least to my understanding. I have an old ass GPU. GeForce 710, 1GB VRAM
sand iris
gt 710 is too old to be supported
But yeah, the UI shows up no matter what. But when it comes to converting audio (inference), if there aren't any supported GPUs, it's unless not going to work, or offload to CPU
meager edge
This computer was handed down to me for free. The CPU is actually not that old but the GPU is old as hell. i5, 8-core, 3.4 Ghz
I appreciate your guys time. I'll invest in a new GPU once I can find one that fits it. This shell is old and restrictive in what I can put into it.
candid olive
is making rvc models legal?
wispy burrow
I appreciate your guys time. I'll invest in a new GPU once I can find one that f...
But you don't need a GPU to be able to do inference.
That works on CPU. Just takes a bit longer.
It's only training that requires an NVIDIA GPU.
meager edge
Oh, okay. I'll still want to invest because this GPU is very limiting. The most "intense" game I can play on here is Rocket League- but even then, I need to lower all the settings completely. I'll still do more research because right now, this interface isn't going to help. I can't adjust audio paths for some reason.
glossy coyote
Oh, okay. I'll still want to invest because this GPU is very limiting. The most ...
lemme know when u have found a solution cuz im on the same boat here 
meager edge
@glossy coyote Gotchu homie
violet heron
is making rvc models legal?
Legal, as long as your not using it for misinformation
rancid knoll
yo
runic thunder
rancid knoll
...whats the problem?
rancid knoll
from my experience you can just ignore it
runic thunder
I use this one
runic thunder
from my experience you can just ignore it
No
Because I cannot download the audio
it doesn't create it
rancid knoll
if it does just try refreshing the page
runic thunder
runic thunder
if it does just try refreshing the page
I will try that
night hazel
Why is this happening even though I have done the preprocessing and extraction?
Traceback (most recent call last):
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 269, in run
utils.latest_checkpoint_path(hps.model_dir, "D_*.pth"), net_d, optim_d
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\infer_libs\train\utils.py", line 212, in latest_checkpoint_path
x = f_list[-1]
IndexError: list index out of range
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 305, in run
torch.load(hps.pretrainD, map_location="cpu")["model"]
File "C:\AI\RVC\Applio-RVC-Fork\runtime\lib\site-packages\torch\serialization.py", line 797, in load
with _open_zipfile_reader(opened_file) as opened_zipfile:
File "C:\AI\RVC\Applio-RVC-Fork\runtime\lib\site-packages\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```After checking I think I might have to use .wav files and not .mp3 files
brittle wing
Uhhh.... where is crepe!??
slow kraken
"Starting Applio with runtime for Nvidia support (you must have it installed).
The system cannot find the path specified.
Press any key to continue"
What am I supposed to do?
azure marshBOT
RVC Guides (How to Make AI Cover)
upbeat ice
hey, so anyone got a simple guide on how to create and train a voice model on hugging face or something?
rare gobletBOT
@candid olive has reached level 1.
slow kraken
"Starting Applio with runtime for Nvidia support (you must have it installed).
T...
Nvm, if anyone's got the same problem click on go-applio-manager.bat
void condor
I have a problem "Train Feature Index" button dont dont anything, even in the console just stay at the last line when the model finished training
oblique venture
Guys what is pitch extraxtion method and crepe hop length?
vernal cypress
i left my pc on the whole night to train a rvc v2 model but when i tried using it, i got this error
Exception: Incorrect format for path\to\file\G_73200.pth. Use a voice model trained using RVC v2 instead.
south tree
ValueError Traceback (most recent call last)
<ipython-input-8-3014b0039645> in <cell line: 18>()
16 import IPython.display as ipd
17 ipd.clear_output()
---> 18 ipd.Audio(opt_path)
/usr/local/lib/python3.10/dist-packages/IPython/lib/display.py in init(self, data, filename, url, embed, rate, autoplay, normalize, element_id)
114 if self.data is not None and not isinstance(self.data, bytes):
115 if rate is None:
--> 116 raise ValueError("rate must be specified when data is a numpy array or list of audio samples.")
117 self.data = Audio._make_wav(data, rate, normalize)
118
ValueError: rate must be specified when data is a numpy array or list of audio samples.
upbeat pine
When I go to clone, COVER appears saying that it was disconnected because I didn't pay for Google colab, is there another way for me to do COVER?
rare gobletBOT
@upbeat pine has reached level 1.
viral moth
I'm having issues with start_http, it opens for a second then closes, opening in admin mode does the same. Anyone know a fix?
pliant crescent
I'm having issues with start_http, it opens for a second then closes, opening in...
Are you sure you're not clicking on the other file called start_https
viral moth
100%
pliant crescent
And do you even have a good GPU
pliant crescent
Also do you have Python installed with path
viral moth
I don't, the video I was watching had it do this first and never mentioned about Python before.
Pretty sure my PC can't handle it anyway, ngl. I have a 1050 Ti.
pliant crescent
Yeah I think you should install Python because RVC the technology behind wokada requires Python
pliant crescent
Pretty sure my PC can't handle it anyway, ngl. I have a 1050 Ti.
Oof
upbeat pine
When I go to clone, COVER appears saying that it was disconnected because I didn't pay for Google colab, is there another way for me to do COVER?
limber tendon
hey guys, i'm trying to get poetry working but i'm getting an "unable to create process using ... \Python38\python.exe -I -W ignore -' error
anyone run into this problem? or does anyone know where to find poetry debug logs?
it just prints a return code 101 error so i have no idea what to look for
stable crest
How do I make models only using google colab?
stable crest
-colab
azure marshBOT
Google Colabs
rare gobletBOT
@stable crest has reached level 1.
slim trellis
Help!
idk how to use applio
it keeps saying errors
IT BETTER NOT BE ONEDRIVE OR ELSE I'LL LOSE IT.
I DON'T KNOW HOW TO GET RID OF IT
I TRIED EVERYTHING ITS TWO HARD
limber tendon
can u improve ai cover gen from the default settings?
or is that as good as it gets
nvm found the readme lol
i have a 1060 and an m1 with the ai chip, could i get better quality on one of those?
sweet thanks
also any software recs? or just rvc-project?
spiral dawn
Hey can someone help me every time I try to do the AI voice and testing it on audacity the riddle always comes out with the audio sounding horrible and glitchy can someone help me with this?
brittle wing
anyone know what is this TTS? there's a Emotion voice feature at it
please gimme a answer!
proud elbow
i left my pc on the whole night to train a rvc v2 model but when i tried using i...
that file is a training snaphot, use yourmodel.pth in \weights folder instead
vernal cypress
oh i see
rare gobletBOT
@vernal cypress has reached level 1.
vernal cypress
thanks!
glossy coyote
it keeps saying errors
Do you have the runtime folder inside your Applio RVC folder?
If you don't have a folder called runtime, you need to download it. If you do have the folder, you have fix this manually.
glossy coyote
<@456226577798135808> open the manager and download the runtime from there inste...
If you don't have the runtime folder, download it from the manager
rare gobletBOT
@glossy coyote has reached level 3.
stable crest
i need help to fix this
Exception Traceback (most recent call last)
<ipython-input-20-fdc82d4e321e> in <cell line: 48>()
47
48 if not os.path.exists(dataset_path):
---> 49 raise Exception(f'I can't find {dataset} in {os.path.dirname(dataset_path)}.')
50
51 os.makedirs(final_directory, exist_ok=True)
Exception: I can't find /content/drive/MyDrive/dataset/mew.zip in /content/drive/MyDrive/rvcDisconnected//content/drive/MyDrive/datas
I follow the instruction to the rvc disconnected and google documents carefully and i got this error
brittle wing
How to make a miles morales ai voice on phone??
zinc grove
guys can anyone help me to install rvc train n infer im missing something on my c drive nd im not smart enough right now :((((
legit cannot get past step one
does it have to be on my c drive cause i dont have space so i extracted the folder onto the d drive, do i have to reinstall python onto my c drive????
grim bay
Why this is error?
vagrant gale
https://imgur.com/J7STxqW
can anyone help?
im getting an error at the 2. step of training on the applio rvc fork
proper shale
Why this is error?
Use dio as your pitch extractor
proper shale
https://imgur.com/J7STxqW
can anyone help?
im getting an error at the 2. step of...
Connection error isn't always a big error
Normally just visua
l
hazy hollow
yo I am kinda new to this AI stuff but does anyone know how do I get the file link? like for example:
Monkey Friend -
https://drive.google.com/file/d/18bKdYmznGmBD4LOkmOhg82v-SOAgGpTc/view?usp=sharing
Goala Cola -
https://drive.google.com/file/d/19Tgli-_JvwlFXcT3WKuSXDOjYAcQ8mUB/view?usp=sharing
these are only google drive, how do I get them as a hugging face link?
proper shale
Well, you don't really need a Huggingface link for those unless the fork doesn't support Drive links
(Applio seems to support it just fine though)
hazy hollow
really?
proper shale
Yeah
hazy hollow
so, just paste the link?
proper shale
Yeah
hazy hollow
thanks mate
zinc grove
guys what do i do C:\WINDOWS\System32>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
The system cannot find the path specified.
what have i installed wrong
rare gobletBOT
@zinc grove has reached level 1.
vagrant gale
Connection error isn't always a big error
soooo I should just continiue with the next steps?
proper shale
guys what do i do C:\WINDOWS\System32>runtime\python.exe infer-web.py --pycmd ru...
You didn't add Python to path
While installing
proper shale
soooo I should just continiue with the next steps?
If it says all-feature-done in the cmd tab, yes
vagrant gale
it doesn't tho
rare gobletBOT
@vagrant gale has reached level 2.
proper shale
honestly, i do not recommend using harvest, rvmpe is just faster and is comparable to mangio im every what
in every way*
vagrant gale
ok thanks I will try that as soon as I get home 🙏
proper shale
ok thanks I will try that as soon as I get home 🙏
Np! Glad to have helped you :)
vagrant gale
thanks again :DD
grim bay
Use dio as your pitch extractor
So there's no solution for me to use the mangio-crepe and rmvpe?
zinc grove
You didn't add Python to path
there was nothging about that on the github... and then i used a different tutorial, do u have one cwhere i can see that and strart again?
proper shale
So there's no solution for me to use the mangio-crepe and rmvpe?
Probably
proper shale
there was nothging about that on the github... and then i used a different tutor...
Sadly, not any that I know of. But if you install the program there is an option to add to path, afaik
sand iris
You didn't add Python to path
Uh, you sure ?
This is attempting to run the provided python runtime
And it really seems like they are running it as administrator
grim bay
Probably
So are there some huggingface or Google Colab link that mangio-crepe and rmvpe works?
twilit gulch
just asking,
does this name clarify how many echops it had?
wrong one 1 sec
dio-test_e670_s16080.pth
proper shale
670
twilit gulch
670
that isn't alot
considering the model is one of the cleanest there is out there
anyway as the last guy said
if I have 1 full wav file full of my voicelines I can actually train it with 1 file?
does it have to be a compressed file or can it be just 1 wav
grim bay
I believe RVC-HF works
"An error occurred downloading"
rare gobletBOT
@grim bay has reached level 4.
sand iris
does it have to be a compressed file or can it be just 1 wav
So it depends, some software take a zip containing audio file(s) and others just take your file folder as input
grim bay
How to fix that?
twilit gulch
So it depends, some software take a zip containing audio file(s) and others just...
mangio rvc local thing
sand iris
If it's local, assume you don't have to zip. you can have one or multiple files, it'll be fine
sand iris
mangio rvc local thing
No need to compress anything then
twilit gulch
No need to compress anything then
so just 1 wav file in a folder and we good?
sand iris
Ya
twilit gulch
ty
proper shale
if I have 1 full wav file full of my voicelines I can actually train it with 1 f...
yh
proper shale
"An error occurred downloading"
link is wrong
twilit gulch
batch size per gpu?
twilit gulch
link is wrong
for training what graph do I check on tensor
timid falcon
how do you resume training in rvc disconnected? do you need the D_2333333 and G_2333333 files ?
twilit gulch
how do you resume training in rvc disconnected? do you need the D_2333333 and G_...
ye
glad zealot
how do you resume training in rvc disconnected? do you need the D_2333333 and G_...
you download it and put it back to the same place
timid falcon
it sucks because I can't download them anymore since I got timed out
twilit gulch
its pretty down
I think I overtrained jsut a bit
well bad news
the right one is just the left one but inverted
aren't they both supposed to be down
?
uh?
zinc grove
You didn't add Python to path
i reinstalled pythion to path bit it still doesnt work idk what2do lol thanks for ur help tho
twilit gulch
tf is grad norm g tho
grim bay
"An error occurred downloading"
Help please
lament eagle
Can someone help me with this I dont understand this
all the stuff is literally loaded
and when I clicked convert it gives me error??????????
lament eagle
it's renewed now
glad zealot
WHY THE FUCK ARE YOU USING THE WEBUI
yes lewl
lament eagle
dude I won't fucking install locally
glad zealot
is it the old one?
lament eagle
it's renewed now
glad zealot
gib
grim bay
I cannot downlad the model
Help
lament eagle
are you sure there's no space after the link
fossil glen
night hazel
When I train my dataset with V1, it works perfectly, but when I try to do it with V2 it just errors and gives me this error: ```Process Process-1:
Traceback (most recent call last):
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 269, in run
utils.latest_checkpoint_path(hps.model_dir, "D_*.pth"), net_d, optim_d
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\infer_libs\train\utils.py", line 212, in latest_checkpoint_path
x = f_list[-1]
IndexError: list index out of range
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 305, in run
torch.load(hps.pretrainD, map_location="cpu")["model"]
File "C:\AI\RVC\Applio-RVC-Fork\runtime\lib\site-packages\torch\serialization.py", line 797, in load
with _open_zipfile_reader(opened_file) as opened_zipfile:
File "C:\AI\RVC\Applio-RVC-Fork\runtime\lib\site-packages\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```
But instead of the runtime error thing I'm having size mismatches and stuff
grim bay
are you sure there's no space after the link
No, I'm sure about that
sand iris
When I train my dataset with V1, it works perfectly, but when I try to do it wit...
Model failed to download, see if you can re-install
night hazel
Model failed to download, see if you can re-install
The entire applio or the dataset?
sand iris
Applio, your dataset is most likely fine
We see this error a whole lot on the real time voice changer, it happens when some required files aren't downloaded fully (so the archives are corrupted)
night hazel
We see this error a whole lot on the real time voice changer, it happens when so...
Tried reinstalling it, and I assume it updated as well because now the UI is dark. But I tried it again and it doesn't work. Gives me this error:
Traceback (most recent call last):
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 269, in run
utils.latest_checkpoint_path(hps.model_dir, "D_*.pth"), net_d, optim_d
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\infer_libs\train\utils.py", line 212, in latest_checkpoint_path
x = f_list[-1]
IndexError: list index out of range
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "C:\AI\RVC\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 289, in run
net_g.module.load_state_dict(
File "C:\AI\RVC\Applio-RVC-Fork\runtime\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs768NSFsid:
size mismatch for dec.noise_convs.0.weight: copying a param with shape torch.Size([256, 1, 64]) from checkpoint, the shape in current model is torch.Size([256, 1, 80]).
size mismatch for dec.ups.0.weight_v: copying a param with shape torch.Size([512, 256, 20]) from checkpoint, the shape in current model is torch.Size([512, 256, 16]).
size mismatch for enc_q.pre.weight: copying a param with shape torch.Size([192, 513, 1]) from checkpoint, the shape in current model is torch.Size([192, 1025, 1]).```And the feature extraction was suspiciously fast as well
I'd like to point that out
spiral dawn
Hey can someone help me every time I try to do the AI voice and testing it on au...
Help pleases
rare gobletBOT
@spiral dawn has reached level 1.
sand iris
And the feature extraction was suspiciously fast as well
Seems like it didn't detect your dataset
night hazel
Ok so if I have to export my dataset as .wav files, what format would I have to use?
16 bit? 24 bit? 32 bit? Maybe that could be the reason
sand iris
Any Audio format should work
grim bay
I cannot downlad the model
Help
I have no idea what to do
sand iris
There are parts of the UI where it should display text as you do the feature extraction, did they list the path to all of your audio files ?
timid falcon
how exactly do you merge 2 voice models?
rare gobletBOT
@timid falcon has reached level 1.
grim bay
I still don't know how to fix it
night hazel
Seems like it didn't detect your dataset
and how would it not detect my dataset if I have a folder with WAV files and they're all preprocessed successfully?
sand iris
Oh they are ?
Well, I'm not sure, all I know is that we've had users with the same error recently
And knowing how unstable applio is, they might have broken it
rare gobletBOT
@night hazel has reached level 2.
grim bay
How to fix the connection timeout on huggingface?
night hazel
I switched the models from Applio to Mangio and now I'm getting this error from something called STFT (iirc it's a pitch shifting software)
Note: you can still call torch.view_as_real on the complex output to recover the old return format. (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\SpectralOps.cpp:867.)
return _VF.stft(input, n_fft, hop_length, win_length, window, # type: ignore[attr-defined]
Process Process-1:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "C:\AI\RVC\Mangio-RVC-v23.7.0\train_nsf_sim_cache_sid_load_pretrain.py", line 225, in run
train_and_evaluate(
File "C:\AI\RVC\Mangio-RVC-v23.7.0\train_nsf_sim_cache_sid_load_pretrain.py", line 399, in train_and_evaluate
) = net_g(phone, phone_lengths, pitch, pitchf, spec, spec_lengths, sid)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\parallel\distributed.py", line 1156, in forward
output = self._run_ddp_forward(*inputs, **kwargs)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\parallel\distributed.py", line 1110, in _run_ddp_forward
return module_to_run(*inputs[0], **kwargs[0]) # type: ignore[index]
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\lib\infer_pack\models.py", line 739, in forward
z, m_q, logs_q, y_mask = self.enc_q(y, y_lengths, g=g)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\lib\infer_pack\models.py", line 194, in forward
x = self.pre(x) * x_mask
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\conv.py", line 313, in forward
return self._conv_forward(input, self.weight, self.bias)
File "C:\AI\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\conv.py", line 309, in _conv_forward
return F.conv1d(input, weight, bias, self.stride,
RuntimeError: Given groups=1, weight of size [192, 513, 1], expected input[2, 1025, 188] to have 513 channels, but got 1025 channels instead```
austere echo
y'all where can i find the ai voices
night hazel
y'all where can i find the ai voices
austere echo
no like
rare gobletBOT
@austere echo has reached level 1.
night hazel
thorny mural
huggingface i think
austere echo
https://www.weights.gg/
tysm
thorny mural
Running into an issue where when I turn on high RVC quality on a merged model it sounds like glitched/robotic Glados (which is cool but intended). Anyone know how to fix it or why it happens?
austere echo
if your audio is 20muntes long yes
thorny mural
wdym
it sounds normal when i have set RVC quality to low, but it "autotunes" into a robotic voice at high
proper shale
Is this normal?
Yes because CPU inference
austere echo
do you have a video so i can understand because i don't understand anything
like i usually just put the audio the voice maybe change the octave and i convert
grim bay
if your audio is 20muntes long yes
Only 3 mins
thorny mural
here let me record some real quick... anywhere i should send it to?
proper shale
Running into an issue where when I turn on high RVC quality on a merged model it...
What program/fork are you using
grim bay
Yes because CPU inference
How long does it takes in normal for 3 mins file?
thorny mural
What program/fork are you using
o, my bad. just using the voice changer client, not making models :p
rare gobletBOT
@thorny mural has reached level 1.
proper shale
How long does it takes in normal for 3 mins file?
Probably that long depending on what pitch extractor you're using
rare gobletBOT
@proper shale has reached level 30.
proper shale
o, my bad. just using the voice changer client, not making models :p
That's probably why
austere echo
here let me record some real quick... anywhere i should send it to?
here maybe or in dm
grim bay
Probably that long depending on what pitch extractor you're using
I'm using mangio-crepe
thorny mural
here we go, just got perms (ty btw)
proper shale
I'm using mangio-crepe
Yeah definitely
thorny mural
yeah
austere echo
what software are you using
proper shale
W-Okada
thorny mural
oh, didn't know that's what it was called 💀
grim bay
Probably that long depending on what pitch extractor you're using
I have been waiting for like almost 30 mins
austere echo
oh, didn't know that's what it was called 💀
are you on pc?
proper shale
Can you send your settings @thorny mural?
thorny mural
it's on a merged voice exported as an onnx
proper shale
I have been waiting for like almost 30 mins
Yeah you shouldn't have selected Mangio crepe
proper shale
Just use dio
Much much faster
Probably like 600-700 seconds
Because CPU inference
austere echo
why don't you even use easygui
proper shale
Maybe that's your mic
thorny mural
why don't you even use easygui
i don't know what easygui is :p
grim bay
How can I use mangio-crepe and rmvpe?
grim bay
why don't you even use easygui
Huggingface?
proper shale
How can I use mangio-crepe and rmvpe?
Click on Advanced Settings
Change it to dio
Mangio and rvmpe will be extremely slow on CPU
grim bay
But I want to use it somehow
thorny mural
sorry i literally just started on this stuff yesterday
austere echo
but it's a software using rvc that is super easy to understand
austere echo
sorry i literally just started on this stuff yesterday
search for local easygui on google thn it's a github link
rare gobletBOT
@austere echo has reached level 2.
grim bay
And wait... 1200 seconds
How rmvpe is faster than mangio-crepe???
thorny mural
search for local easygui on google thn it's a github link
found it, ty will try that and see if it fixes stuff
proper shale
How rmvpe is faster than mangio-crepe???
It is faster but it's still slow because it's running on a CPU
proper shale
rvmpe is better 90% of the time
also pls don't use EasyGUI locally
mangio is just much more stable
proper shale
rvmpe is better 90% of the time
mangio-crepe is only 10x better in low hop sizes but it takes ages to process those
- more sensitive to noise
making models on mangio is a big no-no
unless you've cleaned it up correctly
austere echo
but why is easygui bad tho
austere echo
like i never had any problem
proper shale
- mangio fork has much more options
slim trellis
Do you have the runtime folder inside your Applio RVC folder?
oh
rare gobletBOT
@obtuse vortex has reached level 2.
red rock
bro
rare gobletBOT
@red rock has reached level 1.
slim trellis
where is python located?
slim trellis
@glossy coyote can you help me out?
violet heron
You can train for a tad bit longer
vapid gust
I’m at 60 epochs
violet heron
Wow
vapid gust
What?
violet heron
75 maybe?
vapid gust
Shouldn’t I just wait till it overtrains
violet heron
I can just stop the cell
Think so….
vapid gust
Ok
proper shale
Read the guides if you haven't yet
compact wraith
If this is tensorboard or a different program?, not sure what you use
vapid gust
Tensorboard
compact wraith
Tensorboard
Probably you know somepoints of how tensor works, but since most of the guides from the old server been deleted
maybe you can try looking at this one
could be the updated one, but i just dont bother with the tensorboard
vapid gust
Nah ik the main points of tensorboard
So
Use scalars > loss/g/total
smoothing max
ignore outliners
and just wait till the graphstarts to peak up quickly
I’m 232 epochs
violet heron
Wow
mellow token
so what’s the problem
violet heron
Good dataset and training
vapid gust
is it overtraining
violet heron
Not yet
vapid gust
cuz I see a mini spike that’s why
mellow token
nah
violet heron
Close to be
vapid gust
At 4K steps
mellow token
it might go back down, but once it spikes back up without stopping that’s when it’s overtraining
brittle wing
Do anyone know any good ai cover websites that is good for phone ( one that has a community option so I can search for ais?)
vapid gust
So ig I should just wait till i really see a spike up
violet heron
Do anyone know any good ai cover websites that is good for phone ( one that has ...
You can use wordband
Discord bot
vapid gust
Good dataset and training
It’s like 7 minutes of studio recordings
brittle wing
Idk what that bot is
Can it sing audios from a yt vid
And can I invite it to my server
glossy coyote
<@186842613461024768> can you help me out?
With? 😮
brittle wing
https://docs.google.com/document/d/1-1t1iQDST5dVExejFnfyZaF_qUwcgqmYqzGOjryyXrQ/...
I'll try it out
vapid gust
mellow token
let it be
brittle wing
@violet heron hey... Where do I use the bot what channel
violet heron
<@858923165379199007> hey... Where do I use the bot what channel
#1159290666019471421 preferably
Any channels work tho
or DMs
brittle wing
Also can I have a link to invite to my server if I can
violet heron
Also can I have a link to invite to my server if I can
I don’t think you can invite them
Would be cool if you were able to
vapid gust
rare gobletBOT
@vapid gust has reached level 5.
vapid gust
brittle wing
Would be cool if you were able to
U can't :(
violet heron
If you want, it can still train for a bit
vapid gust
Let’s say I go afk
Can I go back to the part where
It doesn’t overtrain
Or do I have to stop the cell when I feel it’s gonna overtrain and use that model
Bcuz I remeber you could just leave it and calculate after what is the best part
brittle wing
is this how to download rvmpe
jaunty nimbus
-colab
azure marshBOT
Google Colabs
neat egret
How can i do text to speech with ai vm's i created/downloaded? i want to make an ai vm video of tts
opaque copper
Is there a way I can get the model's voice to stop cracking so much?
rare gobletBOT
@proud elbow has reached level 6.
brittle wing
Why this is error?
They made a higgingface version?
rare gobletBOT
@brittle wing has reached level 4.
void condor
INFO:torch.nn.parallel.distributed:Reducer buckets have been rebuilt in this iteration. What is that?
rare gobletBOT
@void condor has reached level 1.
sand iris
INFO:torch.nn.parallel.distributed:Reducer buckets have been rebuilt in this ite...
nothing useful or to be worried about
sand iris
They made a higgingface version?
yeah it's a thing now
compact wraith
@sand iris I need some assistance, because im very confused about something
void condor
nothing useful or to be worried about
Can I ask, why the training is too slow? i got several problems when i go to v2, it was easier in v1, can I go to that again?
compact wraith
I'm probably gonna be useless but we'll see
I can just conact you on Dms, because it might be long :d
sand iris
sure
sand iris
Can I ask, why the training is too slow? i got several problems when i go to v2,...
Is it that much slower ?
Some have had issues where training was stupid slow for some reason (~8x slower than normal). If you're sure that's the case, there should be a toggle between v1 and v2
void condor
Yes. yesterday i let like 3 hours and it was on 50 epochs
void condor
Some have had issues where training was stupid slow for some reason (~8x slower ...
How can i fix it, or how can i return to v1?
sand iris
Isn't there a setting for it on the training page ?
vapid gust
which files do i use :
void condor
Isn't there a setting for it on the training page ?
you mean the v1 and the v2 check? just that? never try let me a few minutes
void condor
Some have had issues where training was stupid slow for some reason (~8x slower ...
I just reinstall all over again, but this time i wont install runtime, apparently it works, now is faster like normal
lilac adder
newbie here, can we upload our own models to google collab? or have to use the ones listed there?
rare gobletBOT
@lilac adder has reached level 1.
sand iris
Interesting, I want to take a look into it
#1159949278270193734 has a list
brittle wing
Ah I see
compact wraith
newbie here, can we upload our own models to google collab? or have to use the o...
Yes, you have to connect your google drive
and upload your zip model file onto your drive
or if you have the link to it, do that as well
ur using AICoverGen?
void condor
Damn,, new problem i just finish training model but when i click on "train freature index" nothing happens, not even in the console., should i do something else?
azure marshBOT
RVC Guides (How to Make AI Cover)
terse nymph
hi, how do i make vocal inference smoother? when sliding between notes or distortion, it's sort of crackly/autotune-y
it also tends to make it a bit off-key as compared to what the original inference audio was
i'm using applio on AMD
Applio V Mangio(COLAB)
I'd like to obtain that second method's processing, but Applio just.. well, y'know.
plucky locustBOT
Sending you a private login link
north sorrel
Heyo, dumb question. quick search says prefer Safetensor instead of pickle. Is this for image AI only? All the voice models I find says pickle imports detected on huggingface
plucky locustBOT
Community Models
goku (Goku (English Dub))
luffy (Luffy (English Dub))
mickey (Mickey Mouse)
peter (Peter Griffin)
plankton (Plankton)
spongebob (Spongebob)
squidward (Squidward)
waluigi (Waluigi)
... and 200+ more
Check out all of the public community models at https://www.word.band/voice_models
You can use any model using the model id (bolded) in the /cover command. For example:
/cover model:goku youtube:https://www.youtube.com/watch?v=s9FCCiEs84k
To see your private models and register new ones, please start a DM with me!
lilac adder
@compact wraith actually I dont know what I am doing at the moment I am such a noob. I got the voice, got the model but dont know where to use them
plucky locustBOT
You've already made a wordband account!
Your cover for Hazbin Hotel - Happy Day in Hell | Prime Video is ready! We're sending it to you now...
wispy burrow
#1159290666019471421 @normal berry
normal berry
<#1159290666019471421> <@899730208519753779>
My bad, wrong channel
proper shale
Heyo, dumb question. quick search says prefer Safetensor instead of pickle. Is t...
Probably just for Image AI
pickle works just fine ig
proper shale
Have you filtered it out with a karaoke filtering model?
spiral dawn
Have you filtered it out with a karaoke filtering model?
No
proper shale
You should
spiral dawn
How?
proper shale
Make sure to register so you get little queues
Then just select "MDX B (karaoke)" in the dropdown
And click on Separate (iirc)
spiral dawn
And click on Separate (iirc)
I tried to put the detech audio from an video but it keeps saying it’s not an audio track
proper shale
What format is it in
spiral dawn
Mp3
spiral dawn
What format is it in
M4A
proper shale
That's why
spiral dawn
That's why
Which format should I set it to?
rare gobletBOT
@spiral dawn has reached level 2.
proper shale
MP3 or WAV
spiral dawn
MP3 or WAV
Thanks it’s working
proper shale
Np
compact wraith
<@607055818596024320> actually I don`t know what I am doing at the moment I am s...
AICoverGen
mystic oar
Hi im having a problem with using the mangio rvc locally when training my own voice model
It looks like it's working fine but when I load my Dataset and click the preprocessing Button it just loads forever, any Ideas how to fix that?
wispy burrow
Hi im having a problem with using the mangio rvc locally when training my own vo...
Does something show up in the cmd?
mystic oar
yes an error message
2023-10-21 22:10:12 | ERROR | root | Task exception was never retrieved
future: <Task finished name='qe3syn4vwq_16' coro=<Queue.process_events() done, defined at D:\atest\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\queueing.py:343> exception=1 validation error for PredictBody
rigid kettle
the audio is working whnei record like the play back i hear my self w the voice but my actual mic is not picking is up ( is it normal for is to say index file not found whne i press a voice)?
radiant laurel
Help?
wispy burrow
Help?
No spaces.
You've named the Model/Experiment 'Female V'.
you can't have spaces.
radiant laurel
OOOOOOOOOOOOOOO ok
rare gobletBOT
@radiant laurel has reached level 3.
wispy burrow
the audio is working whnei record like the play back i hear my self w the voice ...
Are you using W-Okada?
warm forge
-colab
azure marshBOT
Google Colabs
radiant laurel
No spaces.
You've named the Model/Experiment 'Female V'.
you can't have spaces.
is it working or nah??
wispy burrow
is it working or nah??
The cmd usually says more.
radiant laurel
The cmd usually says more.
wispy burrow
Is it stuck at that or is it doing more?
mystic oar
Does something show up in the cmd?
Im not allowed to post Screenshots in here yet. should I just sent you the last bit of the error message?
rare gobletBOT
@mystic oar has reached level 1.
mystic oar
Nvm now im allowed lmao
wispy burrow
Im not allowed to post Screenshots in here yet. should I just sent you the last ...
And now?
mystic oar
radiant laurel
Is it stuck at that or is it doing more?
it havesnt moved in the last few mins
brittle jay
so how would I know a model is overtrained? this is my voice model singing careless whisper
its getting some of the syllables wrong (prolly spelt syllables wrong lol)
does that mean its overtrained?
rigid kettle
Are you using W-Okada?
what is that? would really appriate it if you help me
brittle jay
wow.
rare gobletBOT
@brittle jay has reached level 5.
brittle jay
nevermind.
wispy burrow
what is that? would really appriate it if you help me
What prorgam are you using? W-Okada?
radiant laurel
Ok what is going on
half cove
i didnt even ping your ass petty mfer
fiery shadow
Which the original huggyface I try to add the model AI voice and I try to convert audio file cause it keeps a appear the error
steady ruin
Ok what is going on
what GPU do you have, I can't train locally with mine so it's not uncommon
steady ruin
there's an updated fork called Applio which has an installer that does everything for you, you might be better off using that
radiant laurel
what GPU do you have, I can't train locally with mine so it's not uncommon
3060 mti
humble bluff
keep gettin an error when generating in replay- it says “Module Soundfile Has No Attribute Sound File Runtime Error”
visual pumice
half cove
😭
tame mica
LMFAOOOO
modern snow
¯_(ツ)_/¯
calm scarab
are there any tutorials for rvc2 disconnected (training)?
void condor
I need help i cant create the index file after traninng, dont know what to do 😦
rare gobletBOT
@void condor has reached level 2.
calm scarab
when i train on rvc2 disconnected it gives me the pth file but not added.index. why is that?
void condor
when i train on rvc2 disconnected it gives me the pth file but not added.index. ...
same problem
tame mica
i don't think you need the index but for model maker submission you need it
rare gobletBOT
@tame mica has reached level 25.
vapid gust
when i train on rvc2 disconnected it gives me the pth file but not added.index. ...
you gotta train feature extraction
calm scarab
i did?
Exception Traceback (most recent call last)
<ipython-input-15-9c42c5055659> in <cell line: 30>()
29 listdir_res = list(os.listdir(feature_dir))
30 if len(listdir_res) == 0:
---> 31 raise Exception("No features exist for this model yet. Did you run Feature Extraction?")
32
33 try:
Exception: No features exist for this model yet. Did you run Feature Extraction?
rare gobletBOT
@calm scarab has reached level 1.
calm scarab
this is what i get
tribal badge
-colab
azure marshBOT
Google Colabs
rare gobletBOT
@tribal badge has reached level 1.
frozen spoke
dnai
oblique venture
should i stop trainning or should i train more? it keep going down with out going flat
900 epoch btw
oblique venture
And show again
lol this is way more flat than i thought
oblique venture
7 mins 12 sec i think
oblique venture
didnt realise that i need to ignore the outliner XD
vapid gust
are u at 900epochs
oblique venture
yea
vapid gust
how much epochs did u set
oblique venture
900
oblique venture
should i train another model
oblique venture
Hmm ok
oblique venture
Damn so i trained it way too much
vapid gust
not necessarily
ur graph shows that it still can train more
did you use studio recordings?
raw vocals
oblique venture
did you use studio recordings?
phone recording, and then processed it to be better quality
vapid gust
ahh
ig then ur model can train more if u want
but it looks like it should be good without going more
olive heron
I need help
oblique venture
how long did that take
3-4 hours
oblique venture
I used colab so yea
vapid gust
yeah that’s good
vapid gust
I need help
yh?
oblique venture
what colab
Not webui, just rvc but run on colab
https://colab.research.google.com/drive/1TU-kkQWVf-PLO_hSa2QCMZS1XF5xVHqs?usp=sharing
ember crow
Not webui, just rvc but run on colab
https://colab.research.google.com/drive/1TU...
that is web ui.
gradio is against colab law sadly so
lean owl
How much epochs would i need for an audio file like this (audio is quite loud btw so turn ur volume down)
ember crow
use RVC disconnected in #1159949278270193734 @oblique venture
or use one of the local forks
lean owl
How much epochs would i need for an audio file like this (audio is quite loud bt...
Im curious because each attempt has been awful
ember crow
Im curious because each attempt has been awful
maybe because the BASS BOOSTED EFFECT
oblique venture
use RVC disconnected in <#1159949278270193734> <@571705894711197717>
Lol i used the rvc v2 disconnected on colab
lean owl
maybe because the BASS BOOSTED EFFECT
It could be
But i remember seeing someone make a model with that sample and it was actually decent
Idk could be my settings too
lean owl
Idk i just leave everything at default
round fog
How much epochs would i need for an audio file like this (audio is quite loud bt...
try increasing the gain by 10
lean owl
try increasing the gain by 10
how do i do that? i use rvc disconnected btw
btw how do u tell if a model is overtrained or no?
sand iris
-overtrain
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
lean owl
oh jesus theres so much in here and it makes it very confusing to understand
ive never looked that deep into it
stoic cloud
oh jesus theres so much in here and it makes it very confusing to understand
If you think about it in a simple way, you just download some software that takes in a lot of audio and outputs audio that sounds like that audio
mystic oar
I just got my Hand on ApplioRvc and it seems to be working Fine.
Unless that it's taking so much time to train my model. The cause is, that it doesnt detect my Gpu ( AMD GPU )
I thought AMD gpus are supported in Applio? Did I do something wrong?
lean owl
If you think about it in a simple way, you just download some software that take...
thats interesting i never knew about that
sand iris
I just got my Hand on ApplioRvc and it seems to be working Fine.
Unless that it...
Training isn't supported for AMD cards
stoic cloud
I just got my Hand on ApplioRvc and it seems to be working Fine.
Unless that it...
AMD GPUs are supported for inference, not training
sand iris
... Shout-out to Applio for being misleading
stoic cloud
That's why I always recommend Nvidia since basically everything supports nvidia
lean owl
my computer came with nvidia
rare gobletBOT
@lean owl has reached level 8.
lean owl
<@888043198902071316> has reached level **8**.
no i didnt
mystic oar
ye I bought my graphics card before that AI stuff got so big T_T
stoic cloud
ye I bought my graphics card before that AI stuff got so big T_T
Yeah, it's unfortunate. Which GPU do you have specifically?
stoic cloud
my computer came with nvidia
Which card?
lean owl
im watching a tutorial about overtraining by jarods journey and its pretty good but jesus its confusing
lean owl
Which card?
idk all i know is it has nvidia
mystic oar
Yeah, it's unfortunate. Which GPU do you have specifically?
Radeon RX 5700 XT
lean owl
getforce rtx? does that count as a card
vagrant gale
can anyone help? I have a gtx 1660 and have been training on applio rvc but the training is very slow... 100 epochs and 25 save frequency
stoic cloud
You should be able to tell in task manager or device manager if you're really bothered about it
stoic cloud
can anyone help? I have a gtx 1660 and have been training on applio rvc but the ...
How many seconds per epoch?
lone nexus
can anyone help me with this?? using applio
Process Process-1:
Traceback (most recent call last):
File "C:\Users\mikel\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 269, in run
utils.latest_checkpoint_path(hps.model_dir, "D_*.pth"), net_d, optim_d
File "C:\Users\mikel\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\lib\infer\infer_libs\train\utils.py", line 212, in latest_checkpoint_path
x = f_list[-1]
IndexError: list index out of range
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "multiprocessing\process.py", line 315, in _bootstrap
File "multiprocessing\process.py", line 108, in run
File "C:\Users\mikel\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\lib\infer\modules\train\train.py", line 289, in run
net_g.module.load_state_dict(
File "C:\Users\mikel\Desktop\Applio-RVC-Fork\Applio-RVC-Fork\runtime\lib\site-packages\torch\nn\modules\module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for SynthesizerTrnMs768NSFsid:
size mismatch for dec.noise_convs.0.weight: copying a param with shape torch.Size([256, 1, 64]) from checkpoint, the shape in current model is torch.Size([256, 1, 80]).
size mismatch for dec.ups.0.weight_v: copying a param with shape torch.Size([512, 256, 20]) from checkpoint, the shape in current model is torch.Size([512, 256, 16]).
size mismatch for enc_q.pre.weight: copying a param with shape torch.Size([192, 513, 1]) from checkpoint, the shape in current model is torch.Size([192, 1025, 1]). ```Pretty good card as far as I know
vagrant gale
How many seconds per epoch?
huh?
vagrant gale
like 1 minute
1-2
then I stopped the process and tried again and it was working worse than before
like 5-6 minutes per epoch
stoic cloud
then I stopped the process and tried again and it was working worse than before
How many audio files do you have and what is your batch size
stoic cloud
Turn that up
vagrant gale
to what amount?
stoic cloud
Try 6
vagrant gale
alr ty I'll try rn
stoic cloud
If that works, try 9, then 12, then 16,
mystic oar
Pretty good card as far as I know
yeah never had some troubles with it until now 🤓
rare gobletBOT
@mystic oar has reached level 2.
stoic cloud
Keep in mind, if you wanna do AI, go for nvidia
lone nexus
can anyone help me with this?? using applio
``` INFO:Yuvi:Loaded pretrained asse...
helpp
rare gobletBOT
@lone nexus has reached level 1.
stoic cloud
What's the batch size?
vagrant gale
6
stoic cloud
Did you try higher?
stoic cloud
Sure
sand iris
No try lower
6GB VRAM is on the low end for training
If it happens to use more than that, it starts vampiring system memory which is horrendously slow
vagrant gale
6GB VRAM is on the low end for training
i tried 6 but it still takes 3 minutes per epoch
stoic cloud
How many wavs do you have
grave bolt
Hi guys, does anyone speak Portuguese?
vagrant gale
it used to do like 300 in 2 hours
vagrant gale
How many wavs do you have
72
grave bolt
Can anyone use the translator and explain how to do the AI voice?
sand iris
Could also be that random slow training issue but I doubt it
but 1xxx nvidia cards are infamous for training slow because of some choices Nvidia made
vagrant gale
Can anyone use the translator and explain how to do the AI voice?
-rvc
azure marshBOT
Hey, @grave bolt!
👇 Here are some useful resources to help you learn how to make ai covers
rare gobletBOT
@vagrant gale has reached level 3.
vagrant gale
Can anyone use the translator and explain how to do the AI voice?
go to documentation and use translate
24
minutes
grave bolt
I speak Portuguese