Brazil
#✨│ai-help
grave bolt
sand iris
-rvc-br
vagrant gale
oop
sand iris
-rvc br
azure marshBOT
Guias RVC (como fazer cover com IA)
vagrant gale
yuh
it used to train very good
but then I alt f4 the process and it begun working very bad
I even reinstalled applio
but still it doesn't work
sand iris
3min / epoch seems about right to me
vagrant gale
That are open
nope
sand iris
If it trained significantly faster it was bugged
vagrant gale
oh-
vagrant gale
is 250 enough for 24 minutes of audio?
stoic cloud
is 250 enough for 24 minutes of audio?
Yes
vagrant gale
:,)
vagrant gale
That's... Uh.. 12.5 hours
left my pc overnight it did like 325 epochs in 8 hours
brb
stoic cloud
left my pc overnight it did like 325 epochs in 8 hours
Maybe it speeds up or slows down a bit, sometimes epochs take longer or shorter
The average time is what you should aim for when predicting the eta
brittle wing
-rvc
azure marshBOT
RVC Guides (How to Make AI Cover)
vagrant gale
Maybe it speeds up or slows down a bit, sometimes epochs take longer or shorter
also quick question, I have a gree vps with 4gb ram and 4 core cpu, idk if they have a dedicated gpu but does there exist a applio no gui version? cuz im running ubuntu server on the vps
stoic cloud
also quick question, I have a gree vps with 4gb ram and 4 core cpu, idk if they ...
Yes but CPU is slow as fuck so don't
vagrant gale
damn
I used to train on google colab but it doesn't work anymore cuz it disconnects you after a while
stoic cloud
I used to train on google colab but it doesn't work anymore cuz it disconnects y...
Yeah, what's your GPU again?
vagrant gale
gtx 1660
lean owl
do yall miss training models with the gradio ui in colab or do u prefer the disconnected version
proud elbow
batch size 3 for a 6 GB GPU
vagrant gale
do yall miss training models with the gradio ui in colab or do u prefer the disc...
i miss gradio ui
cuz it didn't disconnet u
lean owl
same here
vagrant gale
batch size 3 for a 6 GB GPU
i'll try thanks
lean owl
i miss gradio ui
it was also easy
i hope someone takes the ui with the training section and makes a huggingface space out of it
vagrant gale
it was also easy
it's still easy lol
lean owl
yeah you're right
but not as easy as it originally was
with the gradio ui u could figure it out straight away
unlike rvc disconnected or whatever i found that shit impossible until i read the guide
gentle sundial
Is there a tutorial how to use inference in the easyGUI?
glad zealot
it needs to be a zip file
rare gobletBOT
@brittle wing has reached level 3.
hardy plover
How can be resumed a trained model that was interrupted by a SO crash?
I set this to save the data each 10 minutes so if is possible to train more with Applio...
Or if I can't then at least how to create the .index file 😅
brittle wing
it needs to be a zip file
it's still not working
glad zealot
it's still not working
can you screenshot your dataset folder again?
brittle wing
can you screenshot your dataset folder again?
brittle wing
glad zealot
found it
glad zealot
It needs to be your dataset folder in google drive
for me its /content/drive/MyDrive/dataset
brittle wing
It needs to be your dataset folder in google drive
so this one??
brittle wing
this is the empty one tho
glad zealot
where its highlighted in the picture i sent
brittle wing
yeah so this one
glad zealot
yup
brittle wing
oh so i change the directory?
glad zealot
oh so i change the directory?
yes
brittle wing
ok thx
glad zealot
easy gui being hard weee
brittle wing
is 2:36 hours alot of time for a dataset with around 20 minutes and 250 epochs?
glad zealot
thats waay too long
brittle wing
usually it took around 40 minutes before (5 minute datasets)
glad zealot
10-30 is enough
proper shale
Depends on batch_size
glad zealot
OwO this ui looks nice
proper shale
And what GPU's being used
brittle wing
And what GPU's being used
t4
proper shale
OwO this ui looks nice
Ooo
brittle wing
on google colab
proper shale
on google colab
What batch size
brittle wing
What batch size
idk it doesn't let me change it or see it
brittle wing
is 2:36 hours alot of time for a dataset with around 20 minutes and 250 epochs?
i might have fucked up the math but probably not
proper shale
idk it doesn't let me change it or see it
Oh it's that EasyGUI Colab
brittle wing
it takes around 35 secs for 1 epoch
proper shale
So batch_size 12
brittle wing
Oh it's that EasyGUI Colab
there is no gui
proper shale
I know
brittle wing
i mean there is but it's optional and it's banned anyway
proper shale
But the name is EasyGUI hah
glad zealot
there is no gui
im trying to bring it back real quick
proper shale
👀
glad zealot
just need to test if it dies in 10 mins
brittle wing
But the name is EasyGUI hah
it's easy👨
proper shale
i need that applio colab to retrain a model 😭
glad zealot
just need to make sure i dont reload
proper shale
it's easy👨
...easy guy
brittle wing
OH
glad zealot
i need that applio colab to retrain a model 😭
in a bit
proper shale
brittle wing
then yh it is easygui
proper shale
yeah it's probably normal
glad zealot
wee its training
proper shale
i got 100 epochs in 1hr and a few mins in batch_size 8
brittle wing
wee its training
what colab is this
glad zealot
i use 14
glad zealot
what colab is this
the one you sent me but modified lol
brittle wing
oh
glad zealot
error fck
brittle wing
there was a version of easygui i used before gradio got banned and it used to take 40 minutes to train a model
proper shale
there was a version of easygui i used before gradio got banned and it used to ta...
Well it varies a lot
proper shale
😭
high night
wee its training
Wtf how are you using web UI
glad zealot
Wtf how are you using web UI
cause i can
high night
cause i can
I mean what colab is that
rare gobletBOT
@elder cloak has reached level 1.
glad zealot
I mean what colab is that
its the easy gui one
high night
its the easy gui one
Wit modification send link
fathom crater
high night
I mean ipynb
glad zealot
Wit modification send link
nah goodluck making it :D
fathom crater
nah goodluck making it :D
hey can u help me
high night
nah goodluck making it :D
fathom crater
how can i make a model then
glad zealot
try using rvcv2 disconnected
fathom crater
the rvc software on collab doesnt work
glad zealot
-colab
azure marshBOT
Google Colabs
glad zealot
the 2nd link
bleak yacht
How to use a specific language pack from the langs folder? For example, my system is in English and I want to change the language of the interface without changing the system language, how do I do it?
elder cloak
Is this for me too? ;-;
glad zealot
Is this for me too? ;-;
ye you could use it too
its easier than easy gui... at least for me..
elder cloak
Ok thanks!
glad zealot
How to use a specific language pack from the langs folder? For example, my syste...
no idea.. i only use english anywyas
high night
How to use a specific language pack from the langs folder? For example, my syste...
I might have an answer to that
Imma ask my techy friend
bleak yacht
thanks
oblique venture
Guys are there anyway to filter multiple vocal from a song, some songs i wanted to cover got so many voice overlay that it make my model voice feel like it want to die 😅
proper shale
Guys are there anyway to filter multiple vocal from a song, some songs i wanted ...
You could try MedleyVox
oblique venture
You could try MedleyVox
Ty will look into it
rare gobletBOT
@oblique venture has reached level 6.
fathom crater
how many epochs does a thing take
silent oasis
oblique venture
how many epochs does a thing take
Depend on how much dataset you want to train it with
fathom crater
Depend on how much dataset you want to train it with
how can i know because i have 42
oblique venture
how can i know because i have 42
42 mins? Or seconds
oblique venture
in the dataset are 42 files
Then combine the time of all those dataset and calculate it
fathom crater
how tho
oblique venture
Bro u just need to look for how long the audio file last, of every single file, then combine it 💀
fathom crater
oh
sand iris
oh its a similar error to the one i got on eay gui
tfw the models are corrupted
fathom crater
378 epochs god damm
sand iris
tfw the models are corrupted
Candidates are : any .pt or .pth files
brittle wing
is 2:36 hours alot of time for a dataset with around 20 minutes and 250 epochs?
it's been 2 hours
this better be worth it
oblique venture
378 epochs god damm
Wait u calculated 378 seconds?
oblique venture
Ok first
The number of second only estimate it, not actually show how much epoch u gonna need
fathom crater
ik
rare gobletBOT
@fathom crater has reached level 3.
fathom crater
i calculated
that all the files are gonna be 378 epochs
not secounds
after 378 it should be done
oblique venture
that all the files are gonna be 378 epochs
Wait how do you calculate it anyway
torpid valley
guys do you know how to make these glitched voice in real time?
like it re re re re re repeats itself
glad zealot
like it re re re re re repeats itself
did you add the param file yourslef? thats probably why if you did
torpid valley
also im not even sure thats possible
ш вшв ше ыщьурщц
rancid knoll
what?
brittle wing
<:skullfacedistorted:1159654720735035392> it's been 2 hours
wait
which one is it
brittle wing
launch tensorboard
rancid knoll
idk much about colab but there should be a way to start it
vapid gust
reload button on the right > scalars tab > loss/g/total > ignore outliners > smoothing 0.999
watch the graph > when it dips down (good) > when it rises from the stable dip (overtraining) (stop training)
north rivet
will having clips of yelling help my pitch variance in my model or ruin it?
rare gobletBOT
@north rivet has reached level 1.
north rivet
when paired with a more of a reading/conversational tone
i'm using video game voice lines as my dataset
brittle wing
<:skullfacedistorted:1159654720735035392> it's been 2 hours
worth it
rare gobletBOT
@brittle wing has reached level 4.
brittle wing
definetly worth the wait
low canyon
guys what happened to the english plankton cover???
north rivet
is 2:36 hours alot of time for a dataset with around 20 minutes and 250 epochs?
what gpu do u have
brittle wing
what gpu do u have
i'm not running it locally
north rivet
mine with 10 minutes and 300 epochs took 2.5 hours on my 3080 12gb
north rivet
i'm not running it locally
oh gotcha
rare gobletBOT
@low canyon has reached level 1.
fathom crater
wtf how long does it take to finish the training
hybrid hazel
hello im ask something its ok ?
hybrid hazel
wtf how long does it take to finish the training
how many epoch ?
fathom crater
idk
hybrid hazel
idk?
rare gobletBOT
@hybrid hazel has reached level 1.
sly quail
m
rare gobletBOT
@sly quail has reached level 1.
brittle wing
how do i change epochs?
proper shale
Elaborate
scarlet stratus
Is there a way to make multiple rvc converted covers at once?
elder cloak
I got an error doing "pip install -r requirements.txt" how do I fix it?
It says either
-
Loosen the range of package versions you've specified
-
Remove package versions to allow pip attempt to solve the dependency conflict
Idk what to type for either.
elder cloak
That's what I type?
proper shale
17 seconds voice 1000 epochs ok?
go for it
proper shale
It says either
1. Loosen the range of package versions you've specified
2. Re...
I guess you have to manually install all the packages
because Applio has so many stuff
elder cloak
I guess you have to manually install all the packages
Is there a guide for that?
proper shale
Not really, but should be as simple as doing pip install (modulenamehere) on your cmd
elder cloak
Sorry I'm new to this where do I start with manually doing this.
I did this command on cmd
rare gobletBOT
@elder cloak has reached level 2.
proper shale
Try seeing requirements.txt and copying the names of the modules required
elder cloak
Copy them in cmd?
indigo wagon
Did u end up working this out? I'm having the same issue. sounds perfect out of Okada but in-game people say it is robotic, crackling and sounds like a voice changer.
elder cloak
I found the requirement.txt but what next?
proper shale
I found the requirement.txt but what next?
Now copy the modules names and execute the pip install command
proper shale
Copy those names and do pip install and paste them
elder cloak
In the cmd?
proper shale
Yes
amber jolt
INFO:Revali1:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=27.453, loss_kl=9.000
^^^^ So I for the life of me can't get the same voice as the original, the quality is fine but the tone is off, I've been trying a billion things, and I see this code, is this problematic?
sand iris
Did applio break again
proper shale
INFO:Revali1:loss_disc=nan, loss_gen=nan, loss_fm=nan,loss_mel=27.453, loss_kl=9...
Yup, it isn't generating anything
cobalt oyster
sonic tapir
anyone know any working rvc colabs? cause all the ones I use get disconnected
proper shale
ANy idea what I've done wrong?
Not sure. I've seen another person with the same issue but we couldn't find a fix
brittle wing
Oh my god, someone help me, I want to resume training on the app but I don't know how, I already processed the backup but I don't know how to continue 😭
sand iris
ANy idea what I've done wrong?
Are you using an nvidia GTX GPU ?
amber jolt
0 NVIDIA GeForce GTX 1660 Ti with Max-Q Design
sand iris
Oh yeah, we've seen that happen before to 16xx GPU users
amber jolt
is there a fix ;w;
sand iris
I think the fix is editing one configuration file and changing a line to half = false ... Really not sure about the exact setting name or the file it's supposed to be set in
sonic tapir
anyone know any working rvc colabs? cause all the ones I use get disconnected 😭😭
brittle wing
anyone know any working rvc colabs? cause all the ones I use get disconnected 😭...
applio came back
rare gobletBOT
@brittle wing has reached level 1.
sonic tapir
im on mobile
brittle wing
Oh my god, someone help me, I want to resume training on the app but I don't kno...
omg someone answer me is important...
amber jolt
Just make sure the name, sample rate, pitch guidance, version, and number of cpu processes are the exact same
brittle wing
oop thnks
amber jolt
don't click process data
or feature extraction
just click start training
also, I'm not actually an expert
I just know from experience
fjkdlsj
sonic tapir
does applio work on mobile?
brittle wing
just click start training
& clicked and doesnt work
polar crescent
does anyone know how to reset the settings in Mangio-RVC ?
I changed something and now all audios sound strange
azure marshBOT
RVC Guides (How to Make AI Cover)
amber jolt
I think the fix is editing one configuration file and changing a line to `half =...
['trainset_preprocess_pipeline_print.py', 'D:\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\datasets\', '40000', '8', 'D:\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0/logs/Revali1', 'False']
D:\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\functional.py:641: UserWarning: stft with return_complex=False is deprecated. In a future pytorch release, stft will return complex tensors for all inputs, and return_complex=False will raise an error.
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]
No clue about code but these stand out to me. should I check the .py files listed there?
sand iris
Nah that's the usual output
There's a file that stores the configuration for training, somewhere, but I don't remember where it is at all
amber jolt
Guess I better see if I can hunt it down
wanton swan
How do you use the tts option for Applio, without it say no such file or directory
How do you use the tts option with other rvc models
rare gobletBOT
@wanton swan has reached level 1.
elder cloak
Okay one more question. Why do I hear the orignal audio and how do I remove it?
normal berry
applio came back
where?
rose kindle
at the online method rvcDisconnected could I somehow put the mangio-rvc-fork folder in drive/MyDrive so that it remains saved in case it is interrupted and only change the extraction directory?
wanton swan
How do you use the tts option for Applio, without it say no such file or directory
How do you use the tts option with other rvc models
@craggy bramble
@candid jetty
I need help with the tts option in applio, help me please?
@sly granite
viscid moss
at the online method rvcDisconnected could I somehow put the mangio-rvc-fork fol...
Sure, that's posible
wanton swan
Sure, that's posible
Help pls
wanton swan
Yes
viscid moss
are u using bark or edge tts?
wanton swan
there are options for both I don't know what to use because there is a option for that and vrc models so I don't know what to do with it
viscid moss
there are options for both I don't know what to use because there is a option fo...
Will you use TTS for English or another language?
wanton swan
English
viscid moss
Ok. Have u already downloaded the model?
wanton swan
rare gobletBOT
@wanton swan has reached level 2.
viscid moss
U need something like this
U need to select the voice to infer. For example: EN US EricNeural
wanton swan
How would that bring the voice for the rvc model?
viscid moss
How would that bring the voice for the rvc model?
The model converts the original voice of the Edge TTS
rose kindle
Sure, that's posible
HOW?
rare gobletBOT
@rose kindle has reached level 2.
wanton swan
The model converts the original voice of the Edge TTS
that helped
viscid moss
HOW?
- Mnually moving/downloading
- Coding, but I don't have the code to upload directly to GDrive 😦
wanton swan
?
wanton swan
I know
viscid moss
I think it's the only way for the voice to have "feeling"
wanton swan
Let me try
amber jolt
There's a file that stores the configuration for training, somewhere, but I don'...
Do you know what half = false is replacing exactly?
sand iris
Do you know what half = false is replacing exactly?
Like what does it do ?
It makes a model that uses 32bit data types instead of 16bit ones. For some reason nvidia fucked up the 1xxx series and computing them is unless buggy, or stupidly slow
But larger data types mean transferring more memory around which also has a performance hit...
amber jolt
ah
okay
Still got no clue which file it could be jflsj I mean, I don't know shit about code but I feel like maybe I might be able to pick something out. Or that's just wishful thinking
Looking in the train folder first and blindly scanning
sand iris
Looking in the train folder first and blindly scanning
update: found it
configs folder
all the .json files, replace "fp16_run": true, to "fp16_run": false,
amber jolt
Oho! Got it. Will run a 30 epoch and see how that goes
rare gobletBOT
@amber jolt has reached level 5.
elder cloak
@sand iris can you help me out too? :-:
elder cloak
Okay one more question. Why do I hear the orignal audio and how do I remove it?
@sand iris this ☝️
The original female vocals is still heard in the background
sand iris
I'm on phone so I can't really listen at the moment 😅
elder cloak
Oh okay well I explained it. Is this a problem you dealt with before?
sand iris
Nope, conversion shouldn't allow anything but the voice + eventually glitches from the background track
Really not sure how the original background vocals could stay without being butchered by the conversion
viscid moss
Okay one more question. Why do I hear the orignal audio and how do I remove it?
I think is a separation issue. How did you get the acapella?
elder cloak
It's from a vocal synthesizer like vocaloid so no music at all in the background
elder cloak
Other models? No just this one
rare gobletBOT
@elder cloak has reached level 3.
elder cloak
?
icy fox
so i got the newest rvc and now i see go-web and go-web-dml. What's the difference
sand iris
so i got the newest rvc and now i see go-web and go-web-dml. What's the differen...
first one is for nvidia, 2nd is for anything else on windows
nimble rock
Did u end up working this out? I'm having the same issue. sounds perfect out of ...
Unfortunately no
marsh sierra
so i got the newest rvc and now i see go-web and go-web-dml. What's the differen...
which rvc?
icy fox
the latest
marsh sierra
yeah i need the name as its my first time in the rvc space
do i send the entire log or just errors
which are more then 7 of
sand iris
send 'em all
sand iris
no clue why that's happenning
marsh sierra
oh ok
sand iris
Looked at the code, I really don't get why it's having trouble
marsh sierra
this happens if i run easiergui
C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0>C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\runtime\python.exe EasierGUI.py --pycmd C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\runtime\python.exe --port 7897
Traceback (most recent call last):
File "C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\EasierGUI.py", line 28, in <module>
from my_utils import load_audio, CSVutil
ImportError: cannot import name 'CSVutil' from 'my_utils' (C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0\my_utils.py)
C:\Users\gabri\Downloads\Mangio-RVC-v23.7.0_INFER\Mangio-RVC-v23.7.0>pause
Press any key to continue . . .
fickle ether
I have a problem where I tried to generate my second ai but it said that the Saul Goodman model doesn't exist
torpid fiber
fickle ether
Guess I'll send evidence
astral perch
Will growls screams and laughs mess up the training?
half cove
zenith frost
what is RMVPE
proper shale
what is RMVPE
a pitch extraction method.
zenith frost
a pitch extraction method.
i see thank you
rare gobletBOT
@zenith frost has reached level 1.
proper shale
i see thank you
Np, glad to have helped.
velvet robin
Can you add sugestions for a new turtorial. All I want is a RVC Studio tutorial.
Oops... I should probably say that in #1159516963014451302 .
azure marshBOT
RVC Guides (How to Make AI Cover)
velvet robin
These are good RVC Tutorials but I wanted a RVC Studio Tutorial. I guess I'll just suggest it if that's okay with you.
sonic tapir
does anyone know any rvc google colabs that still work?
I'm on mobile so i can't install locally 😭
wispy burrow
this happens if i run easiergui
Oh no...You've installed easyGUI...
You may have broken your Mangio-RVC by doing so...
Hope it didn't break Mangio for you...
wide parcel
can someone link me that how to train models efficiently guide
sonic tapir
does anyone know any rvc google colabs that still work?
please someone tell me
wispy burrow
please someone tell me
fickle ether
How do i make my ai album?
rare gobletBOT
@fickle ether has reached level 1.
rapid niche
I had the model maker role in the last server, do i need to resubmit a model to get it back or
ancient flower
the little loss dot isn't moving for my model
I'm using this colab link btw https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=Sb5fzhzEXK8X
rare gobletBOT
@ancient flower has reached level 1.
ancient flower
also I can't find the index file for my model in the logs folder
wispy burrow
also I can't find the index file for my model in the logs folder
Probably because you didn't 'Train Feature Index'?
daring seal
Hey does anyone know how to fix the problem with applio when i start it using Runtime (Nvidia support) it shows me "The system cannot find the path specified."
ancient flower
Probably because you didn't 'Train Feature Index'?
I clicked on the play button beforehand
fickle ether
How do i make my ai album?
anyone?
wispy burrow
Hey does anyone know how to fix the problem with applio when i start it using Ru...
Run Go-applio-manager.bat
[3] Download NVDIA Runtime
daring seal
Run Go-applio-manager.bat
[3] Download NVDIA Runtime
ok tysm 😄
rare gobletBOT
@daring seal has reached level 1.
wispy burrow
I clicked on the play button beforehand
Guess it's saved in the 'project-main' folder?
I don't see any mention of 'logs' in that path.
Never used Colab and such myself...
ancient flower
Never used Colab and such myself...
what do you ususally use? It's been a hot minute since I've last trained or used a new voice model so I'm a bit out of the loop for what's typically used for so and so
(im talking before google shutdown the og colab of course)
wispy burrow
what do you ususally use? It's been a hot minute since I've last trained or used...
Mangio-RVC and Applio-RVC are what I'm using.
Locally.
daring seal
i ran into an error while trying to train a model, how do i fix this?
astral jungle
Does anyone here use Google Colab to train models?
native cedar
Someone reopen this plss
tidal summit
ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects
tidal summit
ERROR: Could not build wheels for fairseq, which is required to install pyprojec...
anyone know?
rare gobletBOT
@tidal summit has reached level 2.
fiery shadow
Hi guys , i got some error from huggy face i need to anyone who resolve this error first error i try to add the model its saids theres an error and 2nd error is connection error or something to try to convert to song covers if you find the solution just let me know
rare gobletBOT
@fiery shadow has reached level 1.
brittle jay
how do you dereverb audio please help me.
astral jungle
Hi guys , i got some error from huggy face i need to anyone who resolve this er...
You can try using this link
fiery shadow
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_...
ok thanks i see if i can try thank you so much 🙂
arctic beacon
I am having so much trouble I found a voice model I like but when I input it in it puts an error at my face
fiery shadow
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_...
one more thing how do i used this ?
astral jungle
one more thing how do i used this ?
Click and run all cells
arctic beacon
I am having so much trouble I found a voice model I like but when I input it in ...
I seriously do not understand
astral jungle
one more thing how do i used this ?
When the gradio link appears, click on it
And download the model
fiery shadow
ok thanks for advice
arctic beacon
can anybody help me? I got the link for the character I want but when I put it in it does not work
rare gobletBOT
@arctic beacon has reached level 1.
astral jungle
can anybody help me? I got the link for the character I want but when I put it i...
What's wrong?
arctic beacon
I get this message when I try to put the voice ai model I found in from this server
astral jungle
I get this message when I try to put the voice ai model I found in from this ser...
Which model do you use using the link?
arctic beacon
I found this model which I thought was perfect andddd well when I tried to import it I got that
astral jungle
I found this model which I thought was perfect andddd well when I tried to impor...
Then you can try this Colab link
fiery shadow
it keep me showing the error of hinaaicover i try to cobert many times
i already connected idk why
jolly jewel
how do i use voice models in https://www.weights.gg/ website?
indigo wagon
Unfortunately no
😢
astral jungle
it keep me showing the error of hinaaicover i try to cobert many times
Have you downloaded the model yet?
candid sparrow
hi why does this show up when i try using RVC colab "our runtime has been disconnected due to executing code that is disallowed in our free of charge tier. Colab subsidizes millions of users and prioritizes interactive programming sessions while disallowing some types of usage as outlined in the FAQ. If you believe this message is in error, file an appeal. Please include any relevant context about your usage."
Your compute unit balance is 0. Purchase more
fiery shadow
ok something like this
marsh sierra
Oh no...You've installed easyGUI...
You may have broken your Mangio-RVC by doing...
Sorry for late response
But i will come back after school
brittle wing
hi why does this show up when i try using RVC colab "our runtime has been discon...
Colab absolutely despises the existence of WebUIs for free tier users and they're on a rampage to ban as much notebooks that have tried to bypass it
rare bramble
Can i change the save frequency withoit having to restart all training?
Like can I stop training, chabge the save frequency and then resume from where it left off?
tropic mica
If i want to use this voice changer, do i need to download anything else besides itself? I've seen some youtube guides mentioning i need to download pytorch or python
hallow wren
hello guys, I know it's not related but just checking in case any of you have an idea about real time voice changer app that can be used during phone calls in iphone
brittle wing
started the bat file but it closed automatically and hasnt changed anything
hallow wren
If i want to use this voice changer, do i need to download anything else besides...
This is w-okada voice changer. Best place is to go through the installtion requirement here https://github.com/w-okada/voice-changer/blob/master/README_en.md
rare gobletBOT
@hallow wren has reached level 2.
hallow wren
This is w-okada voice changer. Best place is to go through the installtion requi...
we also have another channel to support this program related questions #🔍│help-w-okada
weak wedge
Hi all, I've made a voice model with Applio, however, I'm having trouble properly saving it.
When I select Save D and G or Save Voice as the "Save type" and then click the "Save model" button then it logs this in the output information:
Save model
Saved without inference model...
This may take a few minutes, please wait...
Model saved successfully
If I select Save All as the "Save type" and then click the "Save Model" button then it logs:
Save model
Model saved successfully
However in both cases the "finished" folder remains empty afterwards.
Are there any known solutions for this? Perhaps the files get saved elsewhere?
vagrant gale
uhh actually I have the same problem as you rn
I was just about to ask the same question lmao
proper shale
It should be in the folder with your models' name in logs, at least
proper shale
@weak wedge I think this applies for you too
proper shale
The usable pth is in weights btw
brittle wing
where did my model save??
Weights folder and Logs folder.
vagrant gale
huh
there's nothing
in the weights theres a lot of files but also I think the right one is "MODEL.pth"
but there is no index file
brittle wing
in the weights theres a lot of files but also I think the right one is "MODEL.pt...
What did you set the name to?
brittle wing
The index file is the "added" one.
vagrant gale
RESZADNIC
rare gobletBOT
@vagrant gale has reached level 4.
vagrant gale
ah thx 🙏
brittle wing
So yes, the first selected one is correct for the model, and the "added" is the index you need. 🙂
vagrant gale
so the index is added_IVF23...
brittle wing
Yep!
brittle wing
Lol all good, it's confusing first time around. Enjoy! :)
(fyi the extra pth's are just the save points [checkpoint intervals] of your model so whatever you set that to (seemingly 25) it saves at each 25 epoch so just ignore those other files unless your model got overtrained in that case try another one of em. Index will work universally with your model at whatever epoch tho :P)
vagrant gale
ahh alr :D
vagrant gale
Lol all good, it's confusing first time around. Enjoy! :)
it used to be way easier with colab lmao
brittle wing
Yeh somewhat, but as you go on you'll find local training to be more controllable and easier to manage the more you get used to it.
lament eagle
Does anyone have a link to SVC colab
proper shale
Why would you need SVC
west wave
Does anyone know how to stop voice models sounding robotic when using them to sing?
Like they sound super robotic
And the old voice model I used has been taken down
jolly jewel
how do i use voice models in https://www.weights.gg/ website?
steady rock
i have many RVC voice models, i look for an desktop application or like web based but it should run on the computer and not some limited cloud stuff. i mean to create like AI covers
to just put an mp3 or wav and the output is also an mp3 or wav
rigid kettle
yes
rare gobletBOT
@rigid kettle has reached level 1.
brittle wing
Do anyone know how I can use my models I downloaded on a website?
edgy sky
i need help, why can't I see TensorBoard when I'm training? it gives me 403 error
glad zealot
i need help, why can't I see TensorBoard when I'm training? it gives me 403 erro...
if its colab then thats normal.. google somehow broke tensorboard
edgy sky
Yeh somewhat, but as you go on you'll find local training to be more controllabl...
how do I know when the model is trained without TensorBoard? i used to look at the curve and visualize the overtraining point...now how do I do that?
edgy sky
if its colab then thats normal.. google somehow broke tensorboard
ah damn. things change so fast... now how do I train visualizing when to stop?
brittle wing
how do I know when the model is trained without TensorBoard? i used to look at t...
You can just look at the total epoch number in the terminal and determine that way.
Wait nvm you're using Colab.
edgy sky
You can just look at the total epoch number in the terminal and determine that w...
but the epoch number doesn't tell me the loss
glad zealot
ah damn. things change so fast... now how do I train visualizing when to stop?
you can right click the 403 page then open iframe source
brittle wing
Google Colab is out of my knowledge because I train locally 😛, good luck with it though :) seems Goblin's got your back
edgy sky
ohhh okay wait let me see
rare gobletBOT
@edgy sky has reached level 1.
edgy sky
it works, thanks! 😄
glad zealot
it doesnt work sometimes btw
weak wedge
<@1081223075565146223> I think this applies for you too
Awesome, thx!
wispy burrow
Whenever I try to train model, It show an error.
What error? What are you using?
Can't help if no info is given...
calm needle
What error? What are you using?
Can't help if no info is given...
I changed the directory, now It's working, thanks!
rigid kettle
What error? What are you using?
Can't help if no info is given...
hey my voice is working like i can hear it but my actual mic dont pick it up, and is it normal for it to say index file nor found, also my file when i start it dont say bat is only says "start_http"
wispy burrow
hey my voice is working like i can hear it but my actual mic dont pick it up, an...
It's better if you go to #🔍│help-w-okada if you need help with W-Okada.
You should get help faster there.
rich knoll
Hi guys, I need some help with Mangio-RVC-Fork.
I got an unexpected inference result. Here are my steps.
- I cloned repo Mangio-RVC-Fork.
- I downloaded the voice model .pth/.index via voice-models channel.
- For input audio, I extracted vocals using UVR5.
- run Mangio-RVC using "sh run.sh" (MacOS).
- in Model Inference tab, I put voice model, input audio, and leave all configs as default.
- press convert button, I got a weird result. Did I miss some step?
bleak lion
how do i download the voice models from weights.gg?
wispy burrow
how do i download the voice models from weights.gg?
Are you on mobile?
bleak lion
yeah
rare gobletBOT
@bleak lion has reached level 1.
wispy burrow
yeah
For some reason the download button is missing on the mobile version of the site.
Try loading the desktop version.
bleak lion
k thx m8
vagrant gale
qucik question
im trying to locally install aicovergen
but when I install the python requirements the process get's stuck at this
sturdy sage
vagrant gale
"You might need to provide the dependency resolver with a stricter constrains"
I left my computer on for like uhh
8 hours
ashen glen
anyone help??
fatal: destination path 'Mangio-RVC-Tweaks' already exists and is not an empty directory.
OSError Traceback (most recent call last)
<ipython-input-9-bfb8a4688ddf> in <cell line: 13>()
11 get_ipython().system('git clone -b pr-optimization --single-branch https://github.com/alexlnkp/Mangio-RVC-Tweaks.git')
12 #Rename to keep backwards compatibility with old variants of Disconnected
---> 13 os.rename("/content/Mangio-RVC-Tweaks", "/content/Mangio-RVC-Fork")
14 get_ipython().system('git clone https://github.com/maxrmorrison/torchcrepe.git')
15 get_ipython().system('mv torchcrepe/torchcrepe Mangio-RVC-Fork/')
OSError: [Errno 39] Directory not empty: '/content/Mangio-RVC-Tweaks' -> '/content/Mangio-RVC-Fork'
burnt zealot
Whenever I try to train model, It show an error.
if u want help u need to show the error
vagrant gale
but when I install the python requirements the process get's stuck at this
anyone?
proper shale
anyone help??
fatal: destination path 'Mangio-RVC-Tweaks' already exists and is...
make the mangio rvc tweaks folder empty
lone vine
How many epochs should I use for a 3-4 minute dataset?
rare gobletBOT
@lone vine has reached level 1.
proper shale
How many epochs should I use for a 3-4 minute dataset?
There isn't a set amount
You'll have to analyze the graphs
ashen glen
make the mangio rvc tweaks folder empty
where can i find the folder?
proper shale
where can i find the folder?
should be in the colabs file explorer
if ur using colab
ashen glen
thanks i fixed it
rare gobletBOT
@ashen glen has reached level 1.
ashen glen
problem is the gpu
"Exception: No GPU detected; training cannot continue. Please change your runtime type to a GPU."
i need to resume training
proper shale
"Exception: No GPU detected; training cannot continue. Please change your runtim...
woops, gotta change the runtime type
second option on the top right dropdown
ashen glen
its on t4
ashen glen
its usage limits
brittle wing
my res: 50000 ms how can i fix this
rare gobletBOT
@brittle wing has reached level 1.
proper shale
my res: 50000 ms how can i fix this
what's your gpu
proper shale
im using gpu0
But what's your GPU
proper shale
amd radeon rx 6600
what f0 det are you using
brittle wing
what f0 det are you using
what is f0 det
proper shale
the pitch extractors basically...
strange kestrel
every epoch takes 3 minutes 
rare gobletBOT
@strange kestrel has reached level 3.
strange kestrel
thank you bot, very cool
brittle wing
thx :3
strange kestrel
may lord have mercy on my laptop for the rest of the night
ashen glen
every epoch takes 3 minutes <:trkArt:586715907636854806>
seems to be way too fast
oh no nvm
strange kestrel
15 hours to train a model is too fast 
keen remnant
MMVC REAL TIME VOICE CHANGER CLIENT Works on window 7
proper shale
download files has ping
export to onnx
sand iris
MMVC REAL TIME VOICE CHANGER CLIENT Works on window 7
no
brittle wing
just click the export to onnx button
voice changer type is rvc right?
lone vine
Do i need to manually stop training the AI or does it stop automatically?
proper shale
Do i need to manually stop training the AI or does it stop automatically?
After it reaches the total epochs yes
vagrant gale
but when I install the python requirements the process get's stuck at this
anyone?
glossy cipher
Can someone guide me on how to install Tensorboard?
buoyant saddle
uh I need some help with Applio? I am trying to run it (I typed 1 because I have NVIDIA) and is said that I didn't have a path specified?
lone vine
i accidentally did 750 epochs can i manually stop it on like 250 epochs for example?
buoyant saddle
i accidentally did 750 epochs can i manually stop it on like 250 epochs for exam...
not sure. I think you can? But I think it won't have a finished model done.
lone vine
makes sense thank you
buoyant saddle
makes sense thank you
no problem!
azure marshBOT
RVC Guides (How to Make AI Cover)
buoyant saddle
# RVC Guides (How to Make AI Cover)
The applio one doesn't help though in my case.
k.. I am trying to install nvidia runtime and dependicies? let's see how this goes..
rare gobletBOT
@buoyant saddle has reached level 1.
buoyant saddle
<@778990809370198046> has reached level **1**.
bruh I was like level 5 or 10 on the prev server lol
gusty nimbus
does anyone know why on one of the rvc huggingface’s it keeps saying ‘error: none’ when trying to convert?
lone vine
uh I need some help with Applio? I am trying to run it (I typed 1 because I have...
run applio manager
and install nvidia
buoyant saddle
run applio manager
I am, I am installing the nvidia runtime right now, and then the dependicies
lone vine
should work once they are installed
uneven charm
Does anyone have the link to the google colab that actually works?
For inferencing
The one that has a GUI, and is banned by colab
pastel marsh
Hello everyone! Please can anyone help with INFER in RVC, I cant use RVC locally on my old Macbook (Intel), so I need to find a way to use RVC "online", found 2 ways:
- Google Collab Pro - 9.99$ / month
- Paperspace Pro - 8$ / month
So I have few question, it will be amazing if you can help)
- In Google Collab Pro if I will buy "Pro" subscription, can I use RVC (Mangio) 24/7?** for INFER only, not for model making.**
- Will I have any limits in Google Collab Pro, cuz its says I will have 100 compute units and I have no idea will it be enough for INFER, not for model making?
- Question about Paperspace Pro, anyone have experience? Maybe I need to choose Paperspace Pro for INFER not Google Collab Pro?
Thanks a lot!
Have zero skills in Python, but can perfectly follow step by step!)
uneven charm
<#1159949278270193734>
However, AICoverGen requires NVidea gpu even for inferencing.
I am wondering if there is a way to run it locally only using CPU, and I understand it will take longer
proper shale
However, AICoverGen requires NVidea gpu even for inferencing.
I am wondering if ...
I mean... the Colabs listed there work (except maybe AICG-No-WebUI)
uneven charm
But if I use a local runtime for it, it will work
proper shale
Hello everyone! Please can anyone help with INFER in RVC, I **cant** use RVC loc...
...you don't need $9 bucks a month to infer, literally do it for free on RVC-HF
proper shale
But if I use a local runtime for it, it will work
Maybe not for AICoverGen, but I'd say Mangio and Applio will be okay with it
proper shale
No problem! Glad to have helped :)
pastel marsh
...you don't need $9 bucks a month to infer, literally do it for free on RVC-HF
I need it for everyday use, thats why need better speed than RVC-HF, even for Infer.
So do u know or maybe you know someone who can help to make RVC work in Paperspace Pro or Google Collab Pro with subscription?)
Thanks for help!
rare gobletBOT
@pastel marsh has reached level 3.
inland gyro
Usually, how long are the datasets processed? I tried using a 1 min 30secs voice file and it's been processing for over 10 minutes now. (I am running locally)
grand willow
i discovered an error in theeasyGUI, by rejekts #1163883017446629508 message
this day only while extracting the wav files to traIn
tidal summit
anyone know why fairseq wont install
grand willow
it was working well before that
rare gobletBOT
@grand willow has reached level 3.
vapid gust
i discovered an error in theeasyGUI, by rejekts https://discord.com/channels/115...
Be a normal human and use rvc disonectted
grand willow
Be a normal human and use rvc disonectted
what happened to the easyGUI ? it was work like charm
jolly jewel
how do i use the voice models from weights.gg?
rare gobletBOT
@jolly jewel has reached level 1.
dapper ibex
what downloads best for amd ?
viscid moss
how do i use the voice models from weights.gg?
creating an account and downloading, but i recommend to use the ModelAI bot
viscid moss
what downloads best for amd ?
w okada?
jagged briar
-rvc
azure marshBOT
RVC Guides (How to Make AI Cover)
broken quest
I'm trying to run go-web.bat
rare gobletBOT
@broken quest has reached level 1.
broken quest
But I can't get it to do anything
it just hangs on this
G:\RVC-MAngio\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
Matplotlib is building the font cache; this may take a moment.
Traceback (most recent call last):
File "G:\RVC-MAngio\Mangio-RVC-v23.7.0_INFER_TRAIN\Mangio-RVC-v23.7.0\infer-web.py", line 79, in <module>
conn = sqlite3.connect('TEMP/db:cachedb?mode=memory&cache=shared', check_same_thread=False)
sqlite3.OperationalError: unable to open database file
light orchid
anyone know the batch size to gpu ratio?
potent root
tough light
The Google collab I use has this error
"The tensorboard extension is already loaded. To reload it, use:
%reload_ext tensorboard
ERROR: Failed to launch TensorBoard (exited with 1).
Contents of stderr:
2023-10-24 07:28:20.905880: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2023-10-24 07:28:20.906068: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered"
I have collab+
There's way more than that but it's super long
rare gobletBOT
@tough light has reached level 1.
strong belfry
The Google collab I use has this error
"The tensorboard extension is already lo...
I don’t think you can use tensor flow on colab
rotund burrow
hello i need help which Version i should download for Nvidia
tough light
I don’t think you can use tensor flow on colab
I've been using it before
I used it yesterday
I still have 91 compute units
strong belfry
hello i need help which Version i should download for Nvidia
If local, applio fork should be easiest.
strong belfry
I've been using it before
Sorry then I’m not too sure. You may need to ensure tensorflow isn’t already open as said in the log.
Also is the gpu currently connected?
tough light
Sorry then I’m not too sure. You may need to ensure tensorflow isn’t already ope...
Idk how to check that, it's not in an open tab. Also yes T4 is connected
strong belfry
Is there a terminal you can open up in colab? Try the command to restart like it said. I don’t use colab myself so I’m unfamiliar
rare gobletBOT
@strong belfry has reached level 9.
tough light
Is there a terminal you can open up in colab? Try the command to restart like it...
Not that I know of, it's not like a computer I have access to.
hard gust
I got original RVC to work with AMD GPU through ROCm on Linux.
Is this possible to run faster trainers like Applio or Mangio through ROCm? I couldn't make it work 😦
strong belfry
I got original RVC to work with AMD GPU through ROCm on Linux.
Is this possible ...
I don’t believe the other forks train much faster. Maybe for extracting pitch with rvmpe for inference and processing dataset.
strong belfry
Oh actually I did notice applio using less vram while training
strong belfry
Maybe that’s because it actually unloads the inference model when train starts
Yes
I was able to move from 8 to 16 on a 12Gb card
Although higher batch might not help quality
hard gust
I couldn't make Applio work on my ubunto system, it's so frustrating
I have a 12GB GPU too, 6700 XT
rotund burrow
i have a Nvida Geforce RTX 2080 Ti which one is good
hard gust
right now a 20 min dataset needs like 100 seconds for every epoch
strong belfry
I wouldn’t be able to help with AMD. As my card is 3060
strong belfry
right now a 20 min dataset needs like 100 seconds for every epoch
Batch size?
hard gust
Yeah, the community's knowledge about ROCm and Linux is tiny and it makes sense 😦
hard gust
Batch size?
- It won't let me set it higher than this while using the original RVC.
strong belfry
Hm that seems about 2x slower than my 3060
hard gust
I'll try to figure out how to run Applio on Linux
hard gust
Hm that seems about 2x slower than my 3060
Yeah which makes no sense cuz this GPU is fire
He's like 3070 TI in terms of performance
strong belfry
I still don’t know how the directml applio one is working or not
hard gust
It works for inference only
tight cairn
I wouldn’t be able to help with AMD. As my card is 3060
Ive got an AMD, guy helped me in the OG sever but I dont have that link he shared with me anymore so
I cant be much help
hard gust
I've installed Linux on an external SSD to dual boot it with my Windows just so I can train 👿
I mean, it's a miracle it works even if it's not the fastest
but I'll try to make it better and run Applio
strong belfry
For amd it was significant less painful to just spend money on a cloud host
Before I got a 3060
tight cairn
Who was it with the AMD that needed help?
strong belfry
I think now paperspace monthly sub with the free GPUs can go 6hrs
rare gobletBOT
@tight cairn has reached level 1.
hard gust
Hmmmm I prefer to not pay at all, I've had this SSD lying on the desk with no use
So for me it was perfect
rare gobletBOT
@hard gust has reached level 2.
hard gust
Who was it with the AMD that needed help?
Meee
I want to run Applio through ROCm on Linux
tight cairn
Meee
Well here, you can thank @craggy bramble for this
https://docs.google.com/document/d/199lIEtOIKBf4c51SYQ5pHn08pQpuzsKKeF1NVsCTJjk/edit
strong belfry
Isn’t that the voice changer though
hard gust
it seems like it's for Windows and it won't support training until Pytorch is ready for Windows' ROCm
but thank you anyways ❤️
strong belfry
The day rocm works nicely for all AMD gpu is the day I stop giving Nividia my money
strong belfry
I only bent over for a 3060 cause I found a cheap used one a while ago
Been blasting it with stable diffusion, LLaMA and now RVC
hard gust
buying used Nvidia GPU's is always waaay better than those scammy retail prices
I also almost bought a used 4070ti
but my gaming performance is so good with 6700xt that I really couldn't do it without feeling stupid 😂
strong belfry
Yeah I have a 6700xt in main system as well
hard gust
I bought it 2 years ago when the price was like 630$ (tax inc.) and it still feels like a steal
today I think you can find new ones in like 330$
is it possible that 50 epochs are enough for a 20min dataset? I mean it looks like it on tensorboard but I want to verify .-.
sand iris
Yeah, the community's knowledge about ROCm and Linux is tiny and it makes sense ...
Sup
Need anything ?
strong belfry
is it possible that 50 epochs are enough for a 20min dataset? I mean it looks li...
Prob not.
hard gust
Need anything ?
omgggg yes
you helped me on a different AMD thing a few months ago also, nice
I want to run Applio through ROCm on Linux but it won't work. After installation it says it can't find Python's path but on every other Python thing it works just fine...
sand iris
What's it complaining about exactly ? Applio isn't all that reliable, and mostly only tested on windows
Could be that something is hardcoded
hard gust
I'll copy the exact error, brb
sand iris
👍
hard gust
What's it complaining about exactly ? Applio isn't all *that* reliable, and most...
Traceback (most recent call last):
File "/home/me/Desktop/Applio-RVC-Fork/assets/configs/config.py", line 37, in find_python_executable
result = subprocess.run(["which", "python"], capture_output=True, text=True, check=True)
File "/usr/lib/python3.10/subprocess.py", line 526, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['which', 'python']' returned non-zero exit status 1.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/me/Desktop/Applio-RVC-Fork/infer-web.py", line 11, in <module>
from assets.configs.config import Config
File "/home/me/Desktop/Applio-RVC-Fork/assets/configs/config.py", line 67, in <module>
python_path = find_python_executable()
File "/home/me/Desktop/Applio-RVC-Fork/assets/configs/config.py", line 42, in find_python_executable
raise Exception("Could not find the Python path on Linux.")
Exception: Could not find the Python path on Linux.
Press Enter to continue...
Only happens with Applio
sand iris
Which distro is that on ?
Really poor guess but could it be that there's no python and only python3 ?
hard gust
Yes, but even after editing Applio's go-applio.sh file to python3, it says the same thing...
sand iris
Try editing under assets/configs/config.py
That line specifically, so that it tries to find python3 instead of python
sand iris
Last thing while I'm at it, see if you are actually using the ROCm torch version
hard gust
Yeah I'll verify this rn
rare gobletBOT
@hard gust has reached level 3.
sand iris
Requirements.txt pulls in torch 2.0.0 which overrides any version you might have installed before
But the requirements-amd file doesn't specify any torch version, letting you get whatever you want
hard gust
Yeah it also happened using the original RVC, so first thing I did is to edit this
icy fox
i wonder when rvc v3 will come out
sand iris
¯_(ツ)_/¯
proper shale
I need it for everyday use, thats why need better speed than RVC-HF, even for In...
Colab Pro will work without any new adjustments. Just go on the Applio Colab and it will work just fine.
Paperspace, however, needs to be set up and all that.
glad zealot
who needs pro when you have me UwU
but ye if you dont wanna do any edits to the code just get pro
proper shale
who needs pro when you have me UwU
frfr
glad zealot
or wait till someone managed to leak my version of it lol
proper shale
keep your stuff unleaked 💀
glad zealot
i mean mia has their own version but it uses a different method
hard gust
Happy to share that Applio is running successfuly on my Linux with AMD GPU using ROCm, and it 2x faster than origial RVC training, so I'm happy!
Thanks for the help man! @sand iris
hard gust
50 seconds for every epoch in this case, instead of 2+ minutes before
hard gust
The GPU displayed in the training tab right ?
yup
sand iris
👍
It used not to before, because the display code was... questionnable, and only matched against nvidia card names lol
Even if it didn't list the GPUs it worked if you gave it the index, it was just unintuitive
hard gust
Well good thing tech is going forward 😂
thanks again, I'm going to eat something and celebrate this success
to make tensorboard update faster I need to edit the log interval on the config json files?
sand iris
That should be it
lament eagle
overtrain?
proper shale
nah, probably still going
lament eagle
alright
proper shale
<@622856015444049937>
Make sure the path above the audios dropdown is the same as your audio's path too
astral jungle
<@622856015444049937>
Show me the google Colab link
astral jungle
Im using local
Link?
small zodiac
Make sure the path above the audios dropdown is the same as your audio's path to...
so the path i should put it from audio?
proper shale
so the path i should put it from audio?
yes
astral jungle
what link?
Link google colab
small zodiac
Im not using google colab
im using the local mangio rvc thing
@proper shale It dont work
small zodiac
Link google colab
bro vc?
proper shale
<@622856015444049937> It dont work
has to be the full path
small zodiac
like this?
proper shale
yes
astral jungle
but then it says error
Bro
proper shale
hmmm
astral jungle
but then it says error
proper shale
what GPU do you have
proper shale
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_...
ok that's good but uh
astral jungle
https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_...
Try using this link
small zodiac
what GPU do you have
Intel(R) Graphics 5500
astral jungle
mangio is better lol
I think mango is blocked
small zodiac
and NVIDIA GeForce 840M
cerulean garnet
how to use the pth and index file for ai voice
astral jungle
how to use the pth and index file for ai voice
Your model link is Huggingface, right??
small zodiac
try using dio as your pitch extractor
tried as well, didnt work
it took time then it stopped
astral jungle
but bro can u give colab for just inference
What does that mean? I don't get it
small zodiac
u can put acapellas in AICoverGen right?\
theres no need for putting a whole song
right?
cerulean garnet
So try the link I just sent
now what do i do
rare gobletBOT
@cerulean garnet has reached level 1.
strange kestrel
greetings my dear friends and AI lovers and bla bla bla
so, i'm using applio, training my model, all is well, training is done, i trained the feature index, training done. now when it gets to saving, it says saved successfully but i can't find the saved folder anywhere
cerulean garnet
so, i'm using applio, training my model, all is well, training is done, i traine...
yo, how do i use the ai voicee
strange kestrel
do you have a model ready
cerulean garnet
do you have a model ready
well like, i have the thingy, pth and index
idk what to do, this is my first time lmfao
lament eagle
should I stop
cerulean garnet
replay's pretty easy to use and so far the results i got are good
do i import anything?
strange kestrel
i mean, all the models that I need, i can already find them on the app, i think there are steps to importing your own voice model into the app
cerulean garnet
i mean, all the models that I need, i can already find them on the app, i think ...
so hitler is also there? lmao js askin
cerulean garnet
oh
small zodiac
@astral jungle
rare gobletBOT
@small zodiac has reached level 4.
cerulean garnet
is that the same thing or is that colab thingy
cerulean garnet
whres this from
strange kestrel
the replay app
rare gobletBOT
@strange kestrel has reached level 4.
cerulean garnet
the replay app
oh seems easy lol
strange kestrel
you can make covers with it
naive fossil
last rvc in local link pls
cerulean garnet
you can make covers with it
oh, its like the musicfy website
strange kestrel
something like that
cerulean garnet
something like that
thanks for the help
strange kestrel
if you need help you can use the guide in general, i think it basically covers everything you need
cerulean garnet
if you need help you can use the guide in general, i think it basically covers e...
ty
naive fossil
last rvc in local link pls
.
brittle wing
does anyone know why my output sound is completely silent?
rose kindle
what is the best option for feature/pitch extraction when we want to train? rmvpe or mangio-crepe?
strange kestrel
@high night god i'm dying slowly
so the folder is saved but i only see an index file, isn't there supposed to be a pth file too
naive fossil
last rvc in local link pls
high night
so the folder is saved but i only see an index file, isn't there supposed to be ...
yup
pth and index both
naive fossil
last rvc in local link pls
high night
should be intact
strange kestrel
i only see index
high night
naive fossil
last rvc in local link pls
strange kestrel
i am slowly dying
high night
yea
strange kestrel
@proper shale you got complimented by a chess term, congratulations
naive fossil
last rvc in local link pls
high night
<@622856015444049937> you got complimented by a chess term, congratulations
bro....
strange kestrel
i need help, when i save my model i only see the file index in the folder, no pth file
thanks in advance michael jackson
or michael jordan
or mary jane if that's more your thing
a brazilian michael jackson doesn't sound bad
brittle wing
why is my exported audio completely silent..??
rare gobletBOT
@brittle wing has reached level 1.
strange kestrel
@high night alright is there any other way for me to save the files without having to retrain the whole thing again
naive fossil
last rvc in local link pls
rare gobletBOT
@naive fossil has reached level 3.
strange kestrel
the fuck is a last rvc in local link
and why does that sound like a GTA cheat code
brittle wing
like no matter what i use
naive fossil
the link to download the last rvc in local
brittle wing
the exported audio is literally dead silent
high night
<@749835767819534378> alright is there any other way for me to save the files wi...
idk i just got here
strange kestrel
i have no idea what that is, i am braindead
high night
i have no idea what that is, i am braindead <:trkArt:586715907636854806>
hes basicALLY ASKING WHAT IS THE LATEST RVC LOCAL SOFTWARE
strange kestrel
THEN CNAT HE IDK FIND IT
brittle wing
the link to download the last rvc in local
the local web-ui thing?
strange kestrel
OI
strange kestrel
NO TO WHICH QUESTION
high night
-EVC