no, I left it at default
#✨│ai-help
pastel oak
ehh that does not look normal yea xd
hallow ferry
ehh that does not look normal yea xd
Let me guess, I went crazy with the number of epoch
pastel oak
I dunno
this is how itd normally look like (the blue line is where my OT is)
how do the other graphs look like
the /mel graph and d/total
hallow ferry
pastel oak
to me it looks like you trained your model, then interrupted it, then instead of continuing from XX epoch you restarted from 0 but the colab shows you continued?
idk
hallow ferry
wait what
pastel oak
im no expert maybe someone else knows
pastel oak
All I know is from the docs file about g/total 😂 and a bit about /mel also needing to sink down in trend
ehh id say try to train from scratch again and have tensorboard running the whole time, see if it still works or does some weird stuff
else id offer to train your dataset for you but im leaving home for 2 days, doubt youd want to wait that long
hallow ferry
I would be grateful because I have already used up all the computational units in colab...
pastel oak
Or you can send me the dataset now, ill run it sometime during the night and send you the result in the morning (im from eu)
pastel oak
I would be grateful because I have already used up all the computational units i...
dm me with a google drive link to the dataset or discord
runic stump
Mangio-RVC go-web.bat "unable to open database file"
I've been trying to solve this or find an answer for months, but I cannot seem to figure out what is going on here. Works fine on my laptop, but it's extremely budget and therefore speeds are too low to actually use. Not sure what to do here, any suggestions help, thanks. Used to use Colab where I trained many models and wish to continue using my models locally.
azure marshBOT
RVC Guides (How to Make AI Cover)
rigid furnace
@pastel oak code to prevent disconnection from inactivity on rvc disconnected?
proud elbow
I would be grateful because I have already used up all the computational units i...
use another account and start over (no resuming)
hallow ferry
use another account and start over (no resuming)
I can make only 100 epoch
proud elbow
I can make only 100 epoch
then target it 100 epochs then at next session resume and target 200 epochs, and so on
hallow ferry
I can’t resume it… it start’s over
proud elbow
keep the D and G files, u can manually move to the logs/yourmodel (via the folder icon in the left pane)
sorry I wont explain more with a fuckin mobile discord
proven hill
im not bugging
worthy tulip
Traceback (most recent call last):
File "/content/Ilaria-RVC/app.py", line 1, in <module>
import subprocess, torch, os, traceback, sys, warnings, shutil, numpy as np
ModuleNotFoundError: No module named 'torch'
i keep getting this
proven hill
worthy tulip
https://media.discordapp.net/attachments/1159290139609137264/1219666894672367676...
ur a godsend thank u
i love how u had this ready
proven hill
i love how u had this ready
you dont even know the quantity of my dms and ping for this
worthy tulip
you dont even know the quantity of my dms and ping for this
im sorry 😭
proven hill
im sorry 😭
noooooo whyyy AHAH
worthy tulip
noooooo whyyy AHAH
i dont wanna add to ur pings my bad
worthy tulip
but have u considered adding a note at the bottom of the rvc
sterile mesa
JSONDecodeError Traceback (most recent call last)
<ipython-input-14-850c0ca7544b> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
Help me with that error please
im using the google colab page (https://colab.research.google.com/drive/1Gj6UTf2gicndUW_tVheVhTXIIYpFTYc7?usp=sharing#scrollTo=Sojp1vLTv59G)
split frigate
what does this do?
does it affect the quality of a model?
like the high number u set, the high quality u get?
and can someone provide me the best settings for the best quality u can get as possible
pastel oak
what does this do?
does it affect the quality of a model?
like the high number ...
you will train until you see overtrain on tensorboard. this can be at 200 epochs or 400 or 300 whatever, so here you put in max number 1000 and then stop when the graph shows you trained too much
split frigate
you will train until you see overtrain on tensorboard. this can be at 200 epochs...
why i cant open it?
pastel oak
why i cant open it?
install Microsoft Visual C++ 14.0
idk if its on this page
or google it
proud elbow
community one is free
split frigate
community one is free
but is it the which is require ?
proud elbow
but is it the which is require ?
it contains the VC++ you need, but idk the standalone installer one
split frigate
how can i stop training, it wont if i click on train model
proud elbow
how can i stop training, it wont if i click on train model
only applio has that feature so the only way to do is force closing the console window
split frigate
only applio has that feature so the only way to do is force closing the console ...
it wont cause any error right?
proud elbow
it wont cause any error right?
nothing matters
rare gobletBOT
Ayo? @proud elbow level 27 !!! 
split frigate
nothing matters
tysm
terse turret
can RVC DISCONNECTED be used for covers
half cove
No
split frigate
idk how to replace the file he mention, how?
the normal way looks robotic voice as u
spark olive
what's the best way to use an AI voice currently for a voice changer?
crystal galleon
i cant use my dedicated radeon graphic
azure marshBOT
RVC Guides (How to Make AI Cover)
hybrid shoal
using colab mine will only work for about 10 minutes before i have to redo the process again
any help
grim bay
How can I use Kaggle for training?
hybrid shoal
like the colab stops?
yeah colab stops and says something abt the code not being supported
runtime disconnected due to code being executed that is not allowed in our free of charge tier
rare gobletBOT
Ayo? @hybrid shoal level 1 !!!
brittle wing
you will train until you see overtrain on tensorboard. this can be at 200 epochs...
Can you show visuals of what we should look for?
The graph shows a slope, I don’t understand what I’m looking at lol.
grim bay
Uh, help
low shard
what's the best way to use an AI voice currently for a voice changer?
if you mean use ai voices in realtime voice changer
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
low shard
ask for help in #🔍│help-w-okada
low shard
can RVC DISCONNECTED be used for covers
no its only for making covers
low shard
JSONDecodeError Traceback (most recent call last)
<ipy...
you are using an outdated colab
check the ones in our docs instead https://docs.aihub.wtf
grim bay
How can I use Kaggle for training AI model??
low shard
How can I use Kaggle for training AI model??
maybe @glad zealot should have one
pastel oak
Can you show visuals of what we should look for?
The graph shows a slope, I don...
example from my recent one
the g/total keeps sinking, this is good
You will keep training until it starts rising back or simply not sinking anymore. Then you use the model before it reached the lowest point
In order to evaluate the lowest point, you just have to keep overtraining, I usually do it like: if my lowest point is at 20k steps, ill keep training until 35k maybe 40k to make sure theres not a new lowest point
then i use the voice model thats clsoest to before the ot like 19,7k whatever number it is
glad zealot
maybe <@444684887363026974> should have one
Multinotebook
pastel oak
in the picture you cant really see if its overtrained, because the lowest point is i think somewhere at 26k but its too early to tell
In my case it reached a new lowest point sometime after i took the screenshot
low shard
Multinotebook
I remembered u had a rvcdisconnected kaggle
or can he just import rvcdisconnected on kaggle
Or u had smt else
glad zealot
I remembered u had a rvcdisconnected kaggle
No? But ye can just import it and just remove any google drive and path
low shard
No? But ye can just import it and just remove any google drive and path
didnt u have smt already made on kaggle for rvc training ?
i think i remember wrong then 😭
glad zealot
didnt u have smt already made on kaggle for rvc training ?
Multinotebook...
low shard
my bad
low shard
No? But ye can just import it and just remove any google drive and path
@grim bay
umbral acorn
Hi. Im using Ilaria RVC. My model is downloaded but i cant choose my model in inference cuz it doesn’t show. What should i do?
rare gobletBOT
Ayo? @umbral acorn level 1 !!!
grim bay
No? But ye can just import it and just remove any google drive and path
Is there a simple tutorial for it? I don't really understand
low shard
Hi. Im using Ilaria RVC. My model is downloaded but i cant choose my model in in...
did u click the refresh buttons
glad zealot
Is there a simple tutorial for it? I don't really understand
Naa
umbral acorn
did u click the refresh buttons
Yes
glad zealot
Lemme find my multinotebook
low shard
Yes
try to re download it and show the output after u download it to see if ti says error
brittle wing
example from my recent one
the g/total keeps sinking, this is good
You will ke...
So then, what epoch to train?
For an example of you’re satisfied with 25K?
umbral acorn
Okay thanks
brittle wing
hypothetical speaking
grim bay
Can I use this link for training?
https://www.kaggle.com/code/darkessid/rvc-v2-disconnected-kaggle
glad zealot
Haven't tested it myself so idk
grim bay
https://www.kaggle.com/code/hinabl/multinotebook-ai-public
Is this the link of Multinotebook?
pastel oak
So then, what epoch to train?
For an example of you’re satisfied with 25K?
choose the maximum (1000) epochs so it keeps training
brittle wing
choose the maximum (1000) epochs so it keeps training
So is the goal, to have a consistent straight line?
rare gobletBOT
Ayo? @brittle wing level 5 !!!
spring orchid
@crystal galleon
crystal galleon
ye
pastel oak
So is the goal, to have a consistent straight line?
the goal in the g/total graph is to not keep lowering itself
thin stump
So is the goal, to have a consistent straight line?
you should be monitoring kl, mel, d/total. Not like 24/7 but you're basically investigating your g/total ^^
pastel oak
could say once its flat or rising
brittle wing
Interesting
pastel oak
you should be monitoring kl, mel, d/total. Not like 24/7 but you're basically in...
Keeping it simple for now haha but you can look at those too, while theyre sinking its good, and once g/total reached the lowest you use the model just before it
theres a detailed guide on those other graphs too if you want to read it
glad zealot
https://www.kaggle.com/code/hinabl/multinotebook-ai-public
Is this the link of M...
Yup
brittle wing
theres a detailed guide on those other graphs too if you want to read it
Yeah, I would like to learn more in depth of the graph.
i’m more of a visual learner, do you know any tutorials?
thin stump
Keeping it simple for now haha but you can look at those too, while theyre sinki...
true. I have a anecdote where g keeps dipping (almost too good to be true). Surely enough, kl has risen up by a few thousand steps so I found the overtraining point
tawny sapphire
whats the point of a help channel if i cant even post screenshots there
rare gobletBOT
Ayo? @tawny sapphire level 1 !!!
brittle wing
true. I have a anecdote where g keeps dipping (almost too good to be true). Sure...
During monitoring, when there’s changes, what do I manipulate?
pastel oak
true. I have a anecdote where g keeps dipping (almost too good to be true). Sure...
ill keep that in mind ty
I dont remember if this had any visuals in it or not
pastel oak
whats the point of a help channel if i cant even post screenshots there
you need lvl 1 which you just reached to post pics
thin stump
During monitoring, when there’s changes, what do I manipulate?
It's in the guide too. You just type g/total, d/total, kl, mel
tawny sapphire
you need lvl 1 which you just reached to post pics
😭
for some reason only beatrice works, none of the other voices work at all
umbral acorn
try to re download it and show the output after u download it to see if ti says ...
I changed my model and it doesn’t show anything
low shard
I changed my model and it doesn’t show anything
You arent connected to gpu btw, click the arrow next to rum in the up right and change the runtime type to T4 Gpu
also send the model link lemme see
thin stump
During monitoring, when there’s changes, what do I manipulate?
actually read this one #1213509354343637065 message That guide has been abandoned by Lusbert
alexmurkoff gave him the yap
for being wrong
umbral acorn
You arent connected to gpu btw, click the arrow next to rum in the up right and ...
um…its saying cant connect gpu…do i have to pay for connecting? 🥲
My gpu is intel iris xe. Does that work?
proud elbow
um…its saying cant connect gpu…do i have to pay for connecting? 🥲
My gpu is int...
use another alt account for the colab, no need to waste ur money
umbral acorn
use another alt account for the colab, no need to waste ur money
Its connected now. Thankyou
knotty tinsel
hey on hugging face all my stuff is working but i cant hear myself it used to work but now it doesn't
rare gobletBOT
Ayo? @knotty tinsel level 1 !!!
umbral acorn
Its connected now. Thankyou
But i still cant pick my model 🥲. Still not showing
hybrid shoal
runtime disconnected due to code being executed that is not allowed in our free of charge tier
help with this?
low shard
um…its saying cant connect gpu…do i have to pay for connecting? 🥲
My gpu is int...
In cloud it doesn't matter ur device gpu, its online, you have to wait 24 hours or use another acc
As ur using google gpu
golden moss
sinfer
grim bay
How can I continue to resume the training on Google Colab when it got disconnect?
hybrid shoal
How can I continue to resume the training on Google Colab when it got disconnect...
someone was saying use kaggle but i wouldnt have a clue on how to make one
indigo aspen
How do I get my model maker role back from the previous AI Hub server? because I still haven't gotten it back yet 🥲
proud elbow
How do I get my model maker role back from the previous AI Hub server? because I...
submit your fine quality model again
umbral acorn
So i finished training in rvc but i only received the .index file but not .pth
I checked weights already but its not there.
Did i do something wrong?
rare gobletBOT
Ayo? @umbral acorn level 2 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
still quartz
Hello, does anyone know how to adjust the pitch in Google Colab? I tried entering numbers like -3 and 2 in the pitch field, but the results were weird. I'm using a translator, so the sentence might be weird.
low shard
Hello, does anyone know how to adjust the pitch in Google Colab? I tried enterin...
For pitch, you can't do much other than guessing which is the most similar to the original one by testing with your ear
tight wind
Can i be sure its OT'ing right now?
pastel oak
Can i be sure its OT'ing right now?
if youre at 6k steps id still keep going. how long is your dataset?
tight wind
like 18m
pastel oak
keep going
tight wind
ok
gleaming echo
Hey guys, has anyone encountered this?
RuntimeError: CUDA error: unknown error CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1. Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.
Training locally on OG RVC
pastel oak
Hey guys, has anyone encountered this?
`RuntimeError: CUDA error: unknown error...
whats your gpu
pastel oak
im not sure if this is the problem didnt fully read through:
https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/8965#issuecomment-1530085758
gleaming echo
Will take a look at it, thanks.
vernal jewel
My voice charger only stays in "update settings" and never loads the voice I want to use
gleaming echo
Will take a look at it, thanks.
I've had tons of CUDA errors before, though.
It's so random and I don't know what's causing all the errors
rare gobletBOT
Ayo? @gleaming echo level 1 !!!
gleaming echo
My hardware should be fully compatible
vernal jewel
This voice charger only stays in "update settings" and never loads the voice I want to use
pastel oak
It's so random and I don't know what's causing all the errors
do you have all the nvidia drivers installed for your gpu + cuda?
gleaming echo
do you have all the nvidia drivers installed for your gpu + cuda?
Yes, I do
Maybe I should switch back to NVIDIA Studio Drivers though
I have the Game Ready Drivers now
I've been trying to make it work for more than a week at this point
rare gobletBOT
Ayo? @glass python level 2 !!!
shadow nimbus
what can i do if i can hear myself but my voice is not changing
unreal lichen
Hi, I trained a model last night with a 12-minute data set without noise and clean audio..It was so clean that training was done quickly But the final output is 40% similar to the original sound i used harvest pitch extraction and 200 epochs. Do you think I should train more epochs or is 12 minutes of data set not enough? I want at least 90% similarity.
gleaming echo
Hi, I trained a model last night with a 12-minute data set without noise and cle...
Check Tensorboard, maybe?
And use a better dataset
High quality one
proud elbow
Hi, I trained a model last night with a 12-minute data set without noise and cle...
bad outdated pitch extraction is bad, go learn why rmvpe is better as state-of-the-art
tight wind
if youre at 6k steps id still keep going. how long is your dataset?
what about now. im almost at 2:30. should i save and continue tmrw or is it ot?
nvm its going down
rare gobletBOT
Ayo? @tight wind level 2 !!!
foggy folio
what is the best settings for mdx23c
tight wind
-colab
tight wind
-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
proud elbow
what is the best settings for mdx23c
generally the default settings are fine, though mdx23c in MVSEP is more optimized
proud elbow
just make a free account
foggy folio
what does that do?
heavy arrow
i havent used RVC in 7 months when the new easy use colab was released by that one dude with the kanye bear pfp (i forget bros name)
what should i do to get back into it
i got a powerful enough gpu to not colab that shit now
proud elbow
i havent used RVC in 7 months when the new easy use colab was released by that o...
-links
azure marshBOT
Suggestions for @heavy arrow
heavy arrow
thanks OG 💜
foggy folio
how do you add an agression under 1 like 0.5 on uvr
lament crest
hello, what is dry vocals?
pastel oak
what about now. im almost at 2:30. should i save and continue tmrw or is it ot?
your last OT was just recently, id still keep going 😄
low shard
You mean the realtime voice changer?
worn pebble
Hey guys. new, i am seeking help on slicing data set. I found out, the webui slice the audio into 325 pieces. and almost all of them were not perfect with cut in the mid of sentence
how do i fix it?
proud elbow
Hey guys. new, i am seeking help on slicing data set. I found out, the webui sli...
slicing based on silence threshold? then do Analyze -> label sounds, in Audacity. the effective settings could be found in the guide https://rentry.co/TrainingVoiceModels
worn pebble
thank you
barren heart
Why easy gui (for rvc v2) cant load model file?
proud elbow
thank you
it would also leave reasonably short leading & trailing silences in each sliced audio, so it could help reducing mode collapses in training
steady hill
Hello, when I run "start_http.bat", the console opens for a second and closes immediately, also if I run it as an administrator
barren heart
Why easy gui (for rvc v2) cant load model file?
rare gobletBOT
Ayo? @barren heart level 2 !!!
low shard
that colab is outdated
check the google colabs inside our docs
inside the cloud folder inside the rvc folder
barren heart
that colab is outdated
can I get the link of new rvc google colab?
pine palm
what should be the batch size for gpu
proud elbow
what should be the batch size for gpu
generally 8 is the stable one and it uses 8 gb vram
pine palm
generally 8 is the stable one and it uses 8 gb vram
i have a laptop with 6g rtx 3060
what should i do
proud elbow
i have a laptop with 6g rtx 3060
it'd still run with some performance bottleneck, still no effect to the resulting quality
gleaming echo
Guys, how do I fix this? 
RuntimeError: CUDA error: unknown error CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
My laptop (HP Omen 16 b0265nw):
CPU: Intel Core i7-11800H
RAM: 32 GB
GPU: RTX 3070 Mobile 8GB
rare gobletBOT
Ayo? @gleaming echo level 2 !!!
pine palm
it'd still run with some performance bottleneck, still no effect to the resultin...
no damage to gpu?
rare gobletBOT
Ayo? @pine palm level 2 !!!
barren heart
how to import voice model?
proud elbow
no damage to gpu?
as long the temp and power are normal as observed in msi afterburner
proud elbow
msi ? its rog
it can run in any hardwares, but there are certainly many other tools, such as GPU-Z
pine palm
it can run in any hardwares, but there are certainly many other tools, such as G...
im noob idk
brittle wing
does anyone notice that the same people can sound very different in 2 interviews
Like, the timbre of the voice is just different
This gives me dataset trouble (which source do I pick that is most accurate)
I theorise it has to do with mic placement, what the room looks like and EQ
and the voice they are putting on
cloud rose
Hi, is it possible to run a RVC model entirely on the cloud without requiring a high-end PC and for free?Can I also develop my own voice model at no cost?
brittle wing
Hi, is it possible to run a RVC model entirely on the cloud without requiring a ...
Yes
-colabs
brittle wing
How?
You can use the google colab spaces and hugginface spaces above
They have GPUs available which you can utilise
split frigate
why it says the directory name is invalid:
it is not my first so idk why
if I ignore it and move to the next step, feature extraction, its appear no-feature-todo
naive estuary
hey, when you make any artist sing any song can you run it locally? for example for vocals/instrumental separation im using the uvr5 tool so i was wondering if theres smth similar to that tool for the ai voices ? I´m new to this here literally joined a few minutes ago ^^ i´d also appreciate if someone can me link me some docs that explains everything. i dont even know what rvc means ^^
sinful nexus
-colab
split frigate
why it says the directory name is invalid:
it is not my first so idk why
if I ...
can any1 tell me the fix
this is annoying
rare gobletBOT
Ayo? @brittle wing level 1 !!!
worn pebble
any one use gradio web ui for rvc?
for training models, i use main wisper model but the probloem is that it deletes sample into 325 fragment with cut in the mid of sentence
any tutorial that can reduce the fragmentation and increase the quality the fragment?
there is no parameter adjustment on the ui to help me on that, and mannually to edit every fragment is way too time consuming
hard halo
Hey there, I tried using UVR for making datasets through the entire process, I'm making a singing model (extracting vocals from song,removing reverb, extracting main vocals and seperating adlibs and removing noise) but I still hear small adlibs.
How can I get rid of those?
brittle wing
how can I upload a model to hugginface
naive estuary
Hey there, I tried using UVR for making datasets through the entire process, I'm...
from my experience when im trying to separate vocals im bumping up the settings like overlap, chunk size etc as high as my pc allows. those basically just take more time to process but will be more precise
rare gobletBOT
Ayo? @naive estuary level 1 !!!
hard halo
from my experience when im trying to separate vocals im bumping up the settings ...
I tried using more aggressive settings but got the same results~, Could it be something on my end?
naive estuary
idk what settings you have for the de-reverb, adlibs tho
hm
only thing i can think of then is trying different model. the ones your using might not have enough data but im not an expert in this so idk
theres prob someone more knowledgable in here
light dust
how can i make my own model
azure marshBOT
RVC Guides (How to Make AI Cover)
hasty crow
I have a problem with tensorboard It doesn't open up I'm on Paperspace not locally
The Applio Docs don't seem to help at all
Also running Applio
umbral acorn
what is your GPU
I only have intel irus xe
Im not sure if this setting correct or not
sand iris
Can't train on that GPU unfortunately
4GB of VRAM is too little, and even if it were to load, it would be about as fast as what you're currently seeing
It's training on CPU rn
umbral acorn
damn 
hard halo
only thing i can think of then is trying different model. the ones your using mi...
I've used the following models:
MDX23C
UVR-DeEcho-DeReverb
UVR-BVE-4B_SN-44100-1
UVR-DeNoise
in the following order
hasty crow
I've used the following models:
MDX23C
UVR-DeEcho-DeReverb
UVR-BVE-4B_SN-44100-...
What do you want to remove exactly the ad libs?
or the chain sounds
hard halo
Adlibs themselves, I tried using de noise with aggressive settings but it doesn't remove it
hard halo
I use FL, how can I remove it if it doesn't work with BVE?
This is what I get with aggression on 50 with BVE, Do I just run it over and over again until I can't hear the adlibs?
rare gobletBOT
Ayo? @hard halo level 2 !!!
trim star
Hey guys, guys, if it anyone anyone could coutt help me out I would extremelly
n at to my current working
appreciate it. I'm trying to add rvmpe+ to my current
Applio or RVC_v2
I tried downloading Aesthetic_Applio forked that includes it but
I can't make it work.
I tried downloading the Applio-RECODE that also includes it,
along with the Recode-Runtime folder and dropped in in that
folder and it's trying to pick my own environment instead of the
env folder.
I am fucking losing my mind over this,
snow hazel
I got this error when trying to open UVR
rigid sparrow
I have a error, somebody help?
naive estuary
anyone using rvc locally? any tool recommendations? ty
rigid sparrow
anyone using rvc locally? any tool recommendations? ty
Exactly
naive estuary
i think it looks like pytorch isnt properly installed
rigid sparrow
because it says that there is not enough memory, but I think I freed up 15GB and it still says that it is not enough
rigid sparrow
I use CPU
naive estuary
maybe you need to setup the page file in windows
if you have enough storage space it might be ram issue? which could be fixed by page file
ima google that thing ur using and see for myself
rare gobletBOT
Ayo? @naive estuary level 2 !!!
naive estuary
it seems like its pretty outdated. do they still work on this project?
rigid sparrow
I'm still working on this project with a few models, so what?
rare gobletBOT
Ayo? @rigid sparrow level 2 !!!
brittle wing
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
How can i fix that?
Chunk 384
extra 32768
gpu Nivida Geforce RTX 3060 Laptop gpu (6 gb)
rigid sparrow
Chunk 384
extra 32768
gpu Nivida Geforce RTX 3060 Laptop gpu (6 gb)
I have other software on my computer, if anything
brittle wing
?
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
yes
low shard
how can i make my own model
check our docs https://docs.aihub.wtf
low shard
[Voice Changer] Waiting generate pipeline...
[Voice Changer] Pipeline is not ini...
for wokada ask help in #🔍│help-w-okada or #1192011222023950368
rigid sparrow
it seems like its pretty outdated. do they still work on this project?
Everything works for me now, thanks!
dense marsh
-rvc
brittle wing
hello, how do i fix the glitchy ai voice chorus in rvc?
rare gobletBOT
Ayo? @brittle wing level 1 !!!
soft stratus
When i train this voice in RVC said to me cannot and no file
Can anyone make this video as voice model
And send me it @low shard
cinder canopy
s there a way to listen to wat the models sound like in the search channel?
soft stratus
s there a way to listen to wat the models sound like in the search channel?
Where
Please i need your help RVC
cinder canopy
wat
cinder canopy
with waht
soft stratus
When i train this voice in RVC said to me cannot and no file
To train this as a voice model
brittle wing
When i train this voice in RVC said to me cannot and no file
this is a bad dataset
there is a lot of clipping
soft stratus
this is a bad dataset
What should i do?
half cove
When i train this voice in RVC said to me cannot and no file
if ur doing mp4 using rvc it wont work
half cove
are there any spaces in the file path?
half cove
show me what your file path is
amber zephyr
whats the difference between all of these ?
anybody?
rare gobletBOT
Ayo? @amber zephyr level 4 !!!
thin stump
whats the difference between all of these ?
Ov2 is for 6 minutes or less, RIN_E3 is for long and perfectly clean datasets so ignore it
amber zephyr
?what if i have 10 minutes
thin stump
Use the original
amber zephyr
okay thanks
amber zephyr
Use the original
btw will it still train as fast if i use original
brittle wing
s there a way to listen to wat the models sound like in the search channel?
if you're talm bout #1175430844685484042, if you click on a thread there will be a preview that the person has uploaded
cinder canopy
if you're talm bout <#1175430844685484042>, if you click on a thread there will ...
about #🔍│find-models
thin stump
lol i was going to say something about that. Ov2 is faster with less steps and original will take a bit longer since it doesn't have a checkpoint to go off of like the pretrains
tardy marsh
Hey the voice changer client opens but doesnt display anything. Anyone help?
brittle wing
about <#1163592055830880266>
doesn't look like it
amber zephyr
lol i was going to say something about that. Ov2 is faster with less steps and o...
ohh okay so in general ov2 trains faster, but will original be at least slightly more detailed?
half cove
lol i was going to say something about that. Ov2 is faster with less steps and o...
checkpoint to go off of like the pretrains
what~4
brittle wing
Can I just say, even if you have (made) a good model, the output will sometimes still not sound very good because the destination voice doesn't have the same intonation as the source person. So they will have the same timbre but still sound like someone else
half cove
both use training checkpoints
thin stump
ohh okay so in general ov2 trains faster, but will original be at least slightly...
it's subjective, use a batch size of 8
proven hill
someone need help?
amber zephyr
it's subjective, use a batch size of 8
alright. and last question. i have a 6 minute and 30 second dataset which I plan on using original anyway, will the time to train be under an hour and 30 minutes?
2 hours and 30 minutes
proven hill
alright. and last question. i have a 6 minute and 30 second dataset which I plan...
theres no right estimation on training times
everything is very dependable on lenght, epoch numbers, ...
batch size too
for 6 minutes and 30 seconds i suggest you doing 300 epochs, batch size 8
amber zephyr
okay
proven hill
with colab should take around 90 minutes i suppose
20 epochs per second iirc
i dont have the paper right on hand rn
amber zephyr
thats fine thanks anyway
azure marshBOT
🎙️ Realtime Voice Changer (Local)
proven hill
thats fine thanks anyway
no prob
soft stratus
tbh i dont know what the actual fuck happened here
Can you help me
exotic cedar
Looks like a bug that requires fixing to me
half cove
same
brave garnetBOT
⠀
Google Colabs 
⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.
AICoverGen-NoWebUI
Useful for making covers, doesn't include a UI, by Ardha, by Eddy, Hina and Gdr.
RVC Disconnected
To train new voice models, by Kit Lemonfoot.
EasyGUI
The OG interface, by Rejects.
⠀
Made with love by Ilaria 💖
azure marshBOT
RVC Guides (How to Make AI Cover)
cinder canopy
wat does changing the index do
half cove
wat does changing the index do
Accents
cinder canopy
aite
half cove
Depends
twilit quartz
fluid matrix
talk yo shit 🗣️
hollow vigil
does any1 know what gpu batch size i should use with a 4060 ti 8gb
proud elbow
does any1 know what gpu batch size i should use with a 4060 ti 8gb
8, but without another game/intensive app running
hollow vigil
8, but without another game/intensive app running
what if a game or intensive app is open
how much then
proud elbow
what if a game or intensive app is open
a lightweight unity game might cause minor bottleneck, but I wont suggest running any games, even valorant, apex, pubg, roblox, etc
hollow vigil
a lightweight unity game might cause minor bottleneck, but I wont suggest runnin...
okay also when i’m using feature extraction
is my cpu which is the i9 14900k supposed to be at 100%
and my gpu 4060 ti at 80-90
and ive been sitting here watching it download for like 5 minutes by now
with nothing in the background but firefox and discord open
proud elbow
is my cpu which is the i9 14900k supposed to be at 100%
its only few minutes and supposed to be, wont cause any system instability as long ur using 2-4 cpu threads
hollow vigil
do u mean 2-4 of this
hollow vigil
its only few minutes and supposed to be, wont cause any system instability as lo...
^
hollow vigil
still recommend to use just 4 regardless your cpu
okay i will then thx
hollow vigil
still recommend to use just 4 regardless your cpu
also do u mind telling me what sample rate and pitch extraction algorithm u use?
im usually rmvpe and 48k
proud elbow
also do u mind telling me what sample rate and pitch extraction algorithm u use?
32k or 40k depending on cutout as seen in the audio spectrogram
also mostly rmvpe
proud elbow
u mean this number?
show the spectrogram view
hollow vigil
how do i do that
proud elbow
in the preferences setting
proud elbow
on audacity?
yea it has spectrogram view setting
hollow vigil
yea it has spectrogram view setting
.
what am i supposed to look at to tell which one i should use
hollow vigil
see at what frequency the cutout is
how do i do that sorry i have no clue what im looking at
ohhh i see
hollow vigil
see at what frequency the cutout is
so if it’s around 16khz i go 32k and what ab 40k which khz do i look at
proud elbow
so if it’s around 16khz i go 32k and what ab 40k which khz do i look at
40k corresponds to 20khz cutout
hollow vigil
40k corresponds to 20khz cutout
okay so just to confirm if it’s 16khz 32k and 20khz 40k and the khz number never change?
so it’ll only be 16khz and 20khz no in between
proud elbow
so it’ll only be 16khz and 20khz no in between
if in between I'd still recommend 32k
hollow vigil
if in between I'd still recommend 32k
okay tysm
topaz cloak
ok this is probably a stupid question but like
how do i make my audio sound more like the model and less like the input audio 😭
every time i do it it sounds like the model is doing an impression of the input audio but when i hear other peoples audio it just sounds geniune
is there a step im missing or is it a problem with the models or what
rare gobletBOT
Ayo? @feral marsh level 2 !!!
bitter mango
Hi,
I don't know if it is right to send messages through this channel? But I have a request about two models; Can you prepare audio files of one person being "Tom Hardy" and the other being "Brian Molko 2022"? I couldn't manage to make a sound file. I want help from you.
Thanks in advance.
azure marshBOT
RVC Guides (How to Make AI Cover)
spare bluff
hey guys I have this https://huggingface.co/huggingtweets/narendramodi but when I insert it in the RVC v2 it doesnt work
pine palm
How long should the mofel voice be ?
thin stump
How long should the mofel voice be ?
In here, the server is dead https://docs.aihub.wtf/essentials/how-to-make-voice-models/
pastel oak
I cant pick vb cable in output
did you install VB-Cable (it says CABLE in front of the input/output) or VAC (it says Line 1)
sacred stone
did you install VB-Cable (it says CABLE in front of the input/output) or VAC (it...
i fixed it by reinstaling vb cable
i had it alrdy but i deleted it and reinstalled
pastel oak
i fixed it by reinstaling vb cable
anyway id recommend installing VAC if you have the other, its better in general
rare gobletBOT
Ayo? @sacred stone level 1 !!!
round plank
yo guys
is this only good for singing?
i've trained a 10 minute data set and it just sounds so bad idk why
pastel oak
i've trained a 10 minute data set and it just sounds so bad idk why
depends on how good your dataset is and how well you trained it. tensorboard keeps track of overtraining points etc
if your dataset is speaking then itll only sound good speaking etc.
round plank
depends on how good your dataset is and how well you trained it. tensorboard kee...
How many epochs are usually needed on an average?
pastel oak
💀
proud elbow
i've trained a 10 minute data set and it just sounds so bad idk why
git gud, learn the guides for cleaning dataset
thin stump
💀
lol i picked it up from #1159289738314919936. I thought they wanted to do it for nefarious purposes like politics until he explained the context
pastel oak
How many epochs are usually needed on an average?
is different everytime, most of my 20-25mins datasets take 250-350 epochs but it depends more on steps
idk if shorter dataset means it needs more training or not
pastel oak
lol i picked it up from <#1159289738314919936>. I thought they wanted to do it f...
Thats genius 😂
round plank
is different everytime, most of my 20-25mins datasets take 250-350 epochs but it...
ah ok, appreciate the help. im gonna keep messing around w it
round plank
git gud, learn the guides for cleaning dataset
wym by cleaning dataset
round plank
is different everytime, most of my 20-25mins datasets take 250-350 epochs but it...
one thing I was confused about is, I put the exact same dataset into elevenlabs and it sounded insanely better
rare gobletBOT
Ayo? @round plank level 4 !!!
lament crest
Hello anyone know how to use audacity??
thin stump
wym by cleaning dataset
It's in the guide. It tells you the order to use on MVSEP in that lil chart https://docs.aihub.wtf/rvc/resources/vocal-isolation/
round plank
It's in the guide. It tells you the order to use on MVSEP in that lil chart http...
thanks
proud elbow
one thing I was confused about is, I put the exact same dataset into elevenlabs ...
seems like the dataset lacks variance in intonations, thus the pitch range (monotonous like naratives/regular TTSes)
thin stump
Hello anyone know how to use audacity??
i'm using it rn
lament crest
What is mvsep
lament crest
i'm using it rn
I want to denounce
thin stump
https://mvsep.com/en and noise gate plugin https://www.auburnsounds.com/products/Renegate.html
round plank
seems like the dataset lacks variance in intonations, thus the pitch range (mono...
So it's too static? like I guess there's not enough rise and fall or smthn?
thin stump
Denoise
you also want kiloheart dynamics here https://rentry.org/RVC-dataset-RX-updated. Those are the settings I use for audacity
thick aurora
where can I use rvc models?
rare gobletBOT
Ayo? @thick aurora level 1 !!!
thin stump
Truncating vocal’s plugin too
It's in the guide I linked if you're referring to truncating silences. You export as multiple files. Select a folder > name your filename as .wav or .flac
lament crest
i downloaded plugin, how can I apply it to audacity..?
thin stump
i downloaded plugin, how can I apply it to audacity..?
It's in your effects. Auburn noise gate and Kiloheart Dynamics right?
lament crest
lemme see
is it available to arm mac right?
yuh I saw it, noise gate!
thank you sooo much
I couldnt do this without you
❤️
thin stump
Yw, you have to adjust it like the ones on the guide
lament crest
okay thankss!
thin stump
okay thankss!
and I wasn't being clear on the truncating question. It's called audio labeling if you ctrl + f the guide
lament crest
okay, ill try it!
hasty crow
i have problems running tensorboard on Applio any help?
it runs on localhost but im on a VM
round plank
If you put the "batch sizer per GPU" higher then the amount of VRAM you have, is it bad?
median trench
Where can i find charles bukowski rvc
compact shale
hello everbody
azure marshBOT
RVC Guides (How to Make AI Cover)
wispy abyss
hey guys. I'm looking for talented devs to partner up with and create a new revolutionary AI system. If you're a talented dev/software engineer and would like to partner up on this amazing journey, send me a message 😆
odd shale
hey guys. I'm looking for talented devs to partner up with and create a new revo...
what
AI system of what?
wispy abyss
i dont know if you know what musicfy is, its essentially a text-to-audio Ai software. What i am looking to create is another AI software, something new that's never been done before (or if it has then there is close to zero competition). I have some distributors that can make this thing go viral, I'm just looking for some devs/software engineers to help me create it. If you're interested in partnering up, then send me a private message
split frigate
why it says the directory name is invalid:
it is not my first so idk why
if I ignore it and move to the next step, feature extraction, its appear no-feature-todo
sand iris
why it says the directory name is invalid:
it is not my first so idk why
if I ...
give it the folder containing audio files
not the file itself
Also why the fuck is it running as admin
split frigate
~~Also why the fuck is it running as admin~~
Idk
split frigate
give it the folder containing audio files
Im stupid. Just realize that
sand iris
Nah it happens lol
split frigate
Tysm
odd shale
i dont know if you know what musicfy is, its essentially a text-to-audio Ai soft...
Nah sorry i got no intelect for that.
ocean socket
-colab
cinder canopy
for rvc, what would be the difference in quality between a 300epoch model and 800epoch model?
rare gobletBOT
Ayo? @cinder canopy level 3 !!!
verbal oxide
anyone have gpt sovits colab?
flint schooner
yo I keep getting this bug, whenever I train the model i only get 0% epoch
I thought it was normal and the way its meant to be untill i did a different model and it worked completely different
not sure how this one is different tho
glad zealot
anyone have gpt sovits colab?
It's on the main repo
verbal oxide
okay i found it
rare gobletBOT
Ayo? @verbal oxide level 8 !!!
glad zealot
for rvc, what would be the difference in quality between a 300epoch model and 80...
Depends on the dataset, sometimes it will be better, sometimes it will be worse
glad zealot
okay i found it
Also hallo been a while
worn trellis
You guys know how I can improve this? The vocal sounds a bit electronic-y, it sounds like a robot singing it.
Maybe I can adjust one of the settings here?
glad zealot
Rmvpe
worn trellis
That's it?
worn trellis
The original singer is a female, and the voice I'm using is a female so that's redundant, I assume?
rare gobletBOT
Ayo? @worn trellis level 1 !!!
glad zealot
Depends still cause some have a higher pitch or lower pitch voice than others
worn trellis
I see, I'll try to play with that setting and see if that does the trick
glad zealot
Most of the time it's just the model itself or the pitch that makes a model to not sound good
burnt zealot
is it possible to train to voices at the sametime
glad zealot
is it possible to train to voices at the sametime
Do multiple ?
glad zealot
You don't want the link to appear in the RVC.
Wot? Can you explain it?
oak sedge
Wot? Can you explain it?
Can I send you the print?
glad zealot
Can I send you the print?
Print?
oak sedge
Print?
yeah, from my screen, so you understand better, I'm Brazilian and have horrible English, I'm using the translator 😦
glad zealot
yeah, from my screen, so you understand better, I'm Brazilian and have horrible ...
Sure
oak sedge
ok, i will send you
rare gobletBOT
Ayo? @oak sedge level 1 !!!
glad zealot
ok, i will send you
You can send it here now too
oak sedge
oh! ok
here
now that it has turned red, even when it was golden, nothing appeared
even with cloudflared, the link doesn't work
i don't know why
I've already turned the computer off and on once, I've already switched accounts and it doesn't work 😦
midnight fern
How do i make a Cover on iOS
low shard
that colab is outdated
use instead the updated colabs in our docs https://docs.aihub.wtf
low shard
How do i make a Cover on iOS
mac or phone ?
midnight fern
iPhone
low shard
check the cloud folder inside the rvc folder in our docs https://docs.aihub.wtf
oak sedge
use instead the updated colabs in our docs https://docs.aihub.wtf
oh!!! thank u so much!!!
low shard
yww
sacred wave
which one is the latest colab stable rvc version please share
midnight fern
Im so confused
rare gobletBOT
Ayo? @midnight fern level 3 !!!
burnt zealot
Do multiple ?
YEA
rare gobletBOT
Ayo? @burnt zealot level 3 !!!
burnt zealot
is it possible?
safe token
guys im having problems with step 2 in the colab
ok nvm, theres an updated colab i didnt know about
rare gobletBOT
Ayo? @safe token level 1 !!!
oak sedge
how does it work??? I'm confuse
gleaming echo
Hi guys! Why does my loss values become NAN after some point of training?
I even left my laptop idle for some time.
Training locally on my laptop.
RTX 3070 Mobile (8 GB)
i7-11800H
32 GB RAM
gleaming echo
how does it work??? I'm confuse
There are YT tutorials
oak sedge
how does it work??? I'm confuse
i´m so confuse
oak sedge
There are YT tutorials
ok
compact shale
hello everbody
After pressing the activator, I get a warning like this
how can i solve this
Isn't this a help channel?
low shard
After pressing the activator, I get a warning like this
how can i solve this
send google colab link
wild adder
How do I use a downloaded voice model, instead of the hugginglabs link
Like where do I put it lol
low shard
How do I use a downloaded voice model, instead of the hugginglabs link
check inside the cloud folder inside the rvc folder of our docs https://docs.aihub.wtf there will be google colabs where you can make ai covers
astral harbor
I used RVC many months ago and it was super easy to use, all I would have to do on collab is run it, get a link, and im in and everything worked flawlessly. Why now do we have to run 30 different things until collab eventually times out and stops working? And is there a way to get the older rvc back?
low shard
I used RVC many months ago and it was super easy to use, all I would have to do ...
older rvc got banned by google colab
so nope
astral harbor
dang
low shard
dang
i could help you to direct you to a easy colab based on your needs, you wanna make a model or an ai cover ?
astral harbor
I have a model but I am just looking to do a cover of someone talking with it
low shard
I have a model but I am just looking to do a cover of someone talking with it
oke, this is what i personally use, https://docs.aihub.wtf/rvc/cloud/ilaria-rvc/, could help you as its not that complicated and its good for making ai covers
and has a UI (interface)
the guide explains you everything you need, but if you go against any errors just ask me
compact shale
send google colab link
low shard
https://colab.research.google.com/drive/1KOS-aisdnvmB-IWyqjz2KsT42DF5_SkE?usp=sh...
idk where you got that colab but i just looked at the code and its outdated
one sec lemme send u the updated one
this is the one for making ai rvc voice models
check the guide out
compact shale
check the guide out
Thank you for your help. I did not want to disturb you with this situation, but since I could not reach the current version, I applied here. Thank you again, I will check it.
flint schooner
Hey I trained my model with the online disconnect collab, I did 150/300 epochs and then the GPU runtime ran out, I set the frequency checkpoint at every 50 but can't find the 150 checkpoint, any help?
old oracle
low shard
Thank you for your help. I did not want to disturb you with this situation, but ...
dw yw
foggy folio
-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
verbal basin
ive got an interesting question
ive noticed every pitch extraction alg still changes pronounciation slighly even without index rate. will there be something coming that is 100% accurate?
brittle wing
hy what is epochs ??
latent kettle
hy what is epochs ??
training cycle. exmaple if you read a book for 1 time it is similar to 1 epoch. thats how RVC look at your dataset.
half cove
ive noticed every pitch extraction alg still changes pronounciation slighly even...
nothing is 100%
you can try rmvpe
half cove
hy what is epochs ??
1 run of the dataset
brittle wing
ok ty
true patrol
is there a way to get okada google colab Real Time Voice changer for free? I'm on MAC, and the GPU isnt that good
half cove
is there a way to get okada google colab Real Time Voice changer for free? I'm o...
you might get disconnected
true patrol
could try this <https://colab.research.google.com/github/hinabl/voice-changer-co...
google colab cost too much moeny 😭
half cove
google colab cost too much moeny 😭
free colab might disconnect but it might work
echo jay
If i want to train my model more for lets say a 100 more epochs, when i set up the training do i need to do the prepcorcessing part again?
Im talking of rvc v2 disconnected on google colab
brittle wing
any way to match voices better? i see people like streamers using these funny ones like kai or duke dennis. and im trynna use them but sound entirely different and nothing like the characters? is there a way to fix or anything?
here he says ''After install if nothing opens rerun the program and it should work''
well it did not
the same thing keeps happening when I click on the thing to download/open it, it opens for 0.2 seconds and closes?
azure marshBOT
RVC Guides (How to Make AI Cover)
tropic trail
-colab
obsidian breach
where do i find the download
rare gobletBOT
Ayo? @obsidian breach level 1 !!!
opal depot
Does anyone know if tortisetts can work with rvc morels? So my gf could type and have it come out as a character in a discord vc?
brittle wing
Is there anyone who can help me in voice channels?
rare gobletBOT
Ayo? @brittle wing level 1 !!!
pastel salmon
guys i need help
in google colab i disconnect every 10 min , is there any solution for it or any thing else like google colab that i can use the gpu
merry axle
can somebody help me?
when I try to generate index in applio, it only generates the "trained" one.
desert niche
-local
azure marshBOT
deep anvil
whats the one i can use while im playing a game
spice thorn
hi, is there any model maker?
dim wraith
ashen vale
why is it that anytime i try to use rvmpe it always gives me an error, every other pitch extraction method works 😭
valid herald
hello does someone know how to do the cloning voice stuff
brittle wing
Not sure if this is the right channel to ask but is there a voice changer app that supports custom models? Ive heard about voice.ai but i found some people to claim its a data mining virus. If anyone can confirm that please lmk
pastel oak
just ask
gilded charm
Not sure if this is the right channel to ask but is there a voice changer app th...
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
gilded charm
> This interaction has expired, use the command `/guides realtime` if you wish t...
Use the first
And also, go #🔍│help-w-okada
quartz badge
"File "C:\Users\user\Documents\Applio-3.0.9\Applio-3.0.9\app.py", line 1, in <module>
import gradio as gr
ModuleNotFoundError: No module named 'gradio'" keep getting this error after installing and trying to run applio
azure marshBOT
RVC Guides (How to Make AI Cover)
odd marsh
.
rare gobletBOT
Ayo? @odd marsh level 1 !!!
red carbon
Hey I'm new here, does anyone know how to create custom models for RVC?
odd marsh
im trying to find out too😭
lime otter
keen laurel
can someone help me with this?
i'm trying to train a model, but when i process the data, it says this "ValueError: invalid literal for int() with base 10: '(my file location)' "
how can i solve this?
modern forum
New here, trying to make my first AI cover, which is Cirno singing Heartaches by Al Bowlly (used in It's Just a Burning Memory by The Caretaker), and the song slowed down by -4 semitones, to then once done making Cirno singing well, reverb and add vinyl static to it, like The Caretaker. The problem is that Cirno sings like a male, and I want to sing on her normal, female voice. I am using RVC GUI to try it.
Any way to make her sing Heartaches in her normal, female voice or better please?
In addition, where I can post the screenshot of my settings I'm using on RVC GUI and 2 audios of Al Bowlly singing Heartaches? (one pitched down due to slowing down the song by -4 semitones and the other the same as the first one but +20 pitch to sound like the unslowed, original sample)
rare gobletBOT
Ayo? @modern forum level 1 !!!
modern forum
Ayo? <@595297828600020995> level **1** !!! <:lfg:1159355870119993496>
you shut up too
split frigate
why i cant install/open
tensorboard ?
rare gobletBOT
Ayo? @split frigate level 4 !!!
odd sun
Hey, guys. I just started messing up with RVC using Applio (it looked like the best from the AI Hub docs), what is the best algorithm for acting? I wont be singing or reading a dramatic text like narration, just normal dialogue.
Also, is Applio the best? Is there any other with more features or more configurations? I'm a software dev (C# tho, nothing related to AI), I can deal with more complex stuff if the results are better.
azure marshBOT
RVC Guides (How to Make AI Cover)
gleaming echo
Why does my loss values become NAN after some point of training?
The model also becomes muted in the result of this.
split frigate
in the middle of training, I got a problem that made it stop, and I just pressed train and it started from the beginning, is it normal or is there something I must do
split frigate
late canyon
-rvc
split frigate
does tensorboard important ? i cant download it
and does inference important?
like it has something with the quality of a model
brittle wing
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
low shard
Is there a video tutorial? I barely get what they want me to do
nope
All yt tuts are outdated
read the guides and ask for help in #🔍│help-w-okada or #1192011222023950368
-realtime
azure marshBOT
-realtime
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
azure marshBOT
RVC Guides (How to Make AI Cover)
ionic adderBOT
azure marshBOT
🎙️ Realtime Voice Changer (Online)
barren zephyr
Traceback (most recent call last): File "/usr/local/lib/python3.10/dist-packages/gradio/routes.py", line 523, in run_predict output = await app.get_blocks().process_api( File "/usr/local/lib/python3.10/dist-packages/gradio/blocks.py", line 1437, in process_api result = await self.call_function( File "/usr/local/lib/python3.10/dist-packages/gradio/blocks.py", line 1109, in call_function prediction = await anyio.to_thread.run_sync( File "/usr/local/lib/python3.10/dist-packages/anyio/to_thread.py", line 33, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "/usr/local/lib/python3.10/dist-packages/anyio/_backends/_asyncio.py", line 877, in run_sync_in_worker_thread return await future File "/usr/local/lib/python3.10/dist-packages/anyio/_backends/_asyncio.py", line 807, in run result = context.run(func, *args) File "/usr/local/lib/python3.10/dist-packages/gradio/utils.py", line 865, in wrapper response = f(*args, **kwargs) File "/content/RVC/demo.py", line 362, in <lambda> fn=lambda name: os.listdir(f'assets/weights/{name}') + glob.glob(f'logs/{name.split(".")[0]}/added_*.index'), FileNotFoundError: [Errno 2] No such file or directory: 'assets/weights/Paul Atreides'
i found a problem when i was trying to prepare to download the model
dire vapor
-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
barren zephyr
`Traceback (most recent call last):
File "/usr/local/lib/python3.10/dist-packa...
AND ITS STILL HAPPENING PLEASE HELP ME
rare gobletBOT
Ayo? @barren zephyr level 1 !!!
hot lagoon
whats a best for free gps on paperspace that dose not run out.
I have a vpn so whats a good spot to do it?
brittle wing
Help pls
JSONDecodeError Traceback (most recent call last)
<ipython-input-9-2977f958e2ef> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
I just loade Viktor Tsoi voice model
safe surge
yo guys what do i do if my vocals have voice cracks
half cove
`Traceback (most recent call last):
File "/usr/local/lib/python3.10/dist-packa...
im guessing this is colab?
what colab is this
half cove
JSONDecodeError Traceback (most recent call last)
<ipy...
-colab
half cove
use one listed below
half cove
yo guys what do i do if my vocals have voice cracks
could be the model the input audio or the settings
half cove
whats a best for free gps on paperspace that dose not run out.
I have a vpn so w...
i thought there was only 1 free gpu?
id just use kaggle
safe surge
could be the model the input audio or the settings
maybe its the music
rare gobletBOT
Ayo? @safe surge level 1 !!!
half cove
maybe its the music
dont infer with music
safe surge
vocals*
tawny gust
Hello guys
I want help
I'm setup the rvc
And when I click on LOCAL CREPE FIX
It dosen't work
And when I click on GO WEB
It doesn't work
Any Help?
half cove
use one listed above
tawny gust
No
I'm doing it in my computer
@half cove What python version should I download to use RVC?
Pls reply
half cove
No problem
-links
azure marshBOT
Suggestions for @tawny gust
half cove
<@936673139419664414> What python version should I download to use RVC?
3.9 or 3.10
hot lagoon
id just use kaggle
Kaggle?
rare gobletBOT
Ayo? @hot lagoon level 21 !!!
half cove
i dont know if the notebooks are public
strong fulcrum
Hello, I was wondering what the best way to do tts with (or without) RVC for mac locally was. I heard tortoise, tacotron, vits, etc are good, or even pairing RVC with a standard tts or one of those are good, but I'm not sure what's the best and how exactly to set it up. I'd like to easily use it and be able to use it in a project as well. Could somebody help? Thanks.
azure marshBOT
RVC Guides (How to Make AI Cover)
low shard
Hello, I was wondering what the best way to do tts with (or without) RVC for mac...
I dont know much about mac locally but you could check this out https://docs.aihub.wtf/tts/tts-tools/
tawny gust
hello
I have this problem
C:\RVC\RVC-beta0717>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
Found
Use Language: en_US
Traceback (most recent call last):
File "C:\RVC\RVC-beta0717\infer-web.py", line 1989, in <module>
with gr.Blocks(theme=gr.themes.Soft(), title="Mangio-RVC-Web 💻") as app:
AttributeError: module 'gradio' has no attribute 'themes'
C:\RVC\RVC-beta0717>pause
Press any key to continue . . .
What should I do?
hello
rare gobletBOT
Ayo? @tawny gust level 2 !!!
midnight vapor
tawny gust
I dont understand
brittle wing
Does anyone know a good source where I can find destination audios
brittle wing
Does anyone know a good source where I can find destination audios
wdym by that
dusty root
Does anyone have a solution?
[Voice Changer] post_update_settings ex: [ONNXRuntimeError] : 7 : INVALID_PROTOBUF : Load model from pretrain/rmvpe.onnx failed:Protobuf parsing failed.
soft stratus
I tried with all the voices and it said no file on easy gui @low shard
brittle wing
wdym by that
voice files I can infer my model to
umbral acorn
How many epoch should i put for 17mins dataset?
im using rvc disc
brittle wing
How many epoch should i put for 17mins dataset?
im using rvc disc
400 or more
best fr
umbral acorn
Okay tysm
soft stratus
I tried with all the voices and it said no file on easy gui
sinful nexus
-colab
spring flare
Bro I can't do anything and everything keeps crashing
rare gobletBOT
Ayo? @spring flare level 1 !!!
brittle wing
Does anyone know why my covers sound very artificial?
soft stratus
I tried with all the voices and it said no file
opal crane
Hello Guys, i have a question :
i'm new to this software and i don't really know why my voice vocals is going down and high and also sometimes robotic!
can anyone help me with the configuration
fiery thistle
Yo i need help
so i was on easy gui on colab at step 2 and i put the link in, and in just gave me an error: "JSONDecodeError: Expecting value: line 1 column 1 (char 0)"
iron cobalt
-hugginface
azure marshBOT
-hugginface
proven hill
im not working
brittle wing
tried it
switched google accounts too
ill just keep trying eventually itll work
hopefully.
whoop dee doo
charred tusk
-help
half cove
i thought it was fixed
the problem is that uv pip install is kinda new and messes up sometimes
proven hill
oh
half cove
could have it check if download worked then loop if it didnt
soft stratus
And in easy gui said error when i click download model
I had a problem with the training model voice
mortal lodge
im tryiung to make a voice model and when im trying to train it it terminates itself with these few lines of code, does anyone know whats wrong? thanks
File "/usr/lib/python3.10/multiprocessing/process.py", line 149, in join
res = self._popen.wait(timeout)
File "/usr/lib/python3.10/multiprocessing/popen_fork.py", line 43, in wait
return self.poll(os.WNOHANG if timeout == 0.0 else 0)
File "/usr/lib/python3.10/multiprocessing/popen_fork.py", line 27, in poll
pid, sts = os.waitpid(self.pid, flag)
File "/usr/local/lib/python3.10/dist-packages/torch/utils/data/_utils/signal_handling.py", line 66, in handler
_error_if_any_worker_fails()
RuntimeError: DataLoader worker (pid 12650) is killed by signal: Terminated.
soft stratus
File "/usr/lib/python3.10/multiprocessing/process.py", line 149, in join
res...
Me too i had this problem
mortal lodge
is it something with the model or did we make the same mistake?
rare gobletBOT
Ayo? @mortal lodge level 1 !!!
mortal lodge
Ayo? <@613458976151044140> level **1** !!! <:lfg:1159355870119993496>
please leave me alone
soft stratus
is it something with the model or did we make the same mistake?
I don't know
proven hill
im tryiung to make a voice model and when im trying to train it it terminates it...
what are you using
mortal lodge
RVC v2 Disconnected
mortal lodge
what are you using
on google collab
mortal lodge
then i dont knw
do you know who may be able to help?
mortal lodge
alright thanks
soft stratus
alright thanks
Wait
soft stratus
I had some problem like you
soft stratus
Waiting for <@444684887363026974> to respond to see if they can help me troubles...
And i mean to talk to him to solve it
rare gobletBOT
Ayo? @soft stratus level 8 !!!
soft stratus
Ayo? <@689023384230559754> level **8** !!! <:lfg:1159355870119993496>
PLEASE STOOPPPPPP😡
mortal lodge
Alright, if I figure it out I’ll lyk
rare gobletBOT
Ayo? @mortal lodge level 2 !!!
barren zephyr
what colab is this
EasyGUI
I used easygui and I clicked the download button but always gives me an error
Annoying.
soft stratus
EasyGUI
I used easygui and I clicked the download button but always gives me an...
Me too.... I had the same problem
half cove
EasyGUI
I used easygui and I clicked the download button but always gives me an...
the new easygui or old?
soft stratus
the new easygui or old?
Old
barren zephyr
the new easygui or old?
The new one I think, probably update in March 2024.
AND EVEN THE NEW ONE STILL DOESNT WORK, THE ERROR KEEPS HAPPENING
i dont use the old colab but ITS HAPPENING TO ME
half cove
The new one I think, probably update in March 2024.
AND EVEN THE NEW ONE STILL ...
-colab
half cove
new one listed here
modern surge
-colab
mighty mirage
what does Epoch mean
ionic zephyr
Does anyone know why I get an error when downloading any model?
JSONDecodeError Traceback (most recent call last)
<ipython-input-26-1d7b23896ae4> in <cell line: 31>()
31 if os.path.exists(config_path):
32 # File exists, proceed with creation of creds and client
---> 33 creds = Credentials.from_service_account_file(config_path, scopes=scope)
34 client = gspread.authorize(creds)
35 else:
5 frames
/usr/lib/python3.10/json/decoder.py in raw_decode(self, s, idx)
353 obj, end = self.scan_once(s, idx)
354 except StopIteration as err:
--> 355 raise JSONDecodeError("Expecting value", s, err.value) from None
356 return obj, end
JSONDecodeError: Expecting value: line 1 column 1 (char 0)
empty ridge
```---------------------------------------------------------------------------
J...
i have the samething to
half cove
use newer colab
ionic zephyr
Thanks
half cove
what does Epoch mean
one run of tha dataset
half cove
But in easy gui said error
the new one?
weird
just try another one
like applio or ila rvc
visual turtle
What Fork RVC with rvmpe_gpu and support for training on 2 gpu's exists? (Applio is out (and Applio v2 too)).
Just my Applio v2 that served me for a long time stopped working and I'm thinking of switching to something else.
azure marshBOT
RVC Guides (How to Make AI Cover)
mighty mirage
one run of tha dataset
What do you mean by that? and why is there all these like 300 Epochs and etc, Is there a way to change that in the app or something?
rare gobletBOT
Ayo? @mighty mirage level 1 !!!
half cove
What do you mean by that? and why is there all these like 300 Epochs and etc, Is...
its done when training the model
its when rvc goes over the dataset one time
mighty mirage
oh? how do i train them
mighty mirage
I see, thank you
fringe zephyr
can i enter a path with multiple audio files in it?
or should i consolidate them all into 1 file
(training locally) Found the answer from the search bar ^^
brittle wing
how do i use thishttps://huggingface.co/spaces/Clebersla/RVC_V2_Huggingface_Version
i want to make ai cover
brittle wing
can anyone help me create an ai cover?
soft stratus
the new one?
Yes
soft stratus
File no found
soft stratus
What error
These
half cove
Like the error message
soft stratus
Like the error message
Yes
soft stratus
U were training correct?
No
half cove
Inference?
soft stratus
Inference?
Yes
thin stump
@empty orchid this part here
I'm more of a visual learner soo.. I don't understand any of it. I already have the dataset spliced and zipped ready to go
for anyone reading, how is the path file related in anyway with colab? Because i'm pretty sure this guide was written for local in mind https://docs.aihub.wtf/tts/gpt-sovits/
mild glen
How do I make a ai model pth file into a index. For voice changer ai
rare gobletBOT
Ayo? @mild glen level 2 !!!
subtle loom
I just found the voice blender feature of applio recently. It is pretty interesting but has left me puzzled.
Most .pth models I have areable to be bleded together but some of them cannot.
Anyone know why certain models are not blendable?
soft stratus
Can anyone solve this? @worn zinc
sharp crag
i need help
is it normal for
python src/download_models.py
took so long and no progression bar shown on git windows?
how big is the files?
edit: fixed for some reason after restart it finished in zap
now i have problem with unable to run webui because of gradio
i tried to manually pip install it but no luck
Traceback (most recent call last): File "E:\App\AI Voice Converter\AICoverGen\src\webui.py", line 8, in <module> import gradio as gr ModuleNotFoundError: No module named 'gradio'
hardy oriole
I need help why I can't download model
hardy oriole
outdated colab
THK
brittle wing
what does index files do btw? i never use it and it still sounds good
azure marshBOT
RVC Guides (How to Make AI Cover)
proven hill
someone need help?
dark lark
is the easy gui by rejects working? i try to create modle and get errors at the download step.
File "/content/RVC/demo.py", line 362, in <lambda>
fn=lambda name: os.listdir(f'assets/weights/{name}') + glob.glob(f'logs/{name.split(".")[0]}/added_*.index'),
FileNotFoundError: [Errno 2] No such file or directory: 'assets/weights/RUKNUTTY'
anyone know what this means?
fierce robin
someone need help?
How to use mangio-crepe, since w-okada never gonna support it.
forest sluice
How to use mangio-crepe, since w-okada never gonna support it.
No need to use it
rmvpe/fcpe is enough to cover
hardy oriole
https://huggingface.co/spaces/r3gm/AICoverGen
Why I can't generate ai cover my voice
ember lily
Someone have problem with Ilaria RVC?
rare gobletBOT
Ayo? @ember lily level 1 !!!
ember lily
After the install I click on the button Start but the public url don't run
@proven hill
low shard
After the install I click on the button Start but the public url don't run
What error do u get
umbral acorn
17m data with 600 epoch still sounds like robot
soft stratus
What error do u get
Can you solve this problem
low shard
Can you solve this problem
What google colab are you using
low shard
17m data with 600 epoch still sounds like robot <:boohooh:1176674698629750975>
More epochs ≠ better, use the Tensorboard instead
If you put too much epochs it could overtrain and be even worse
tired vessel
Anyone know why im now seeing this ?
umbral acorn
If you put too much epochs it could overtrain and be even worse
How to know when my model overtrain?
low shard
Anyone know why im now seeing this ?
For wokada issues ask in #🔍│help-w-okada or #1192011222023950368
low shard
How to know when my model overtrain?