why the fuck did the guide not mention that no is the default
#✨│ai-help
brittle wing
i spent 4 hours training -_-
i was following this
omg i wanna kms
I'm on 220/1000 epoch, and this is my tensorboard. Is it great?
How can I upload media?
rare gobletBOT
Ayo? @brittle wing level 1 !!! 
brittle wing
odd shale
How can I make my voice model
Do you want to make a model of yourself?
odd shale
Umm.. It looks fine for now but you should check g/mel and loss/d/total
If one of those two go up, then your model is overtraining
Which was the dataset length you used?
It will depend on the dataset length and the quality of audio.
timber ridge
in there other channel iut was stated 12 minutes
modest delta
Hi guys
how can i create my voice model
@odd shale sorry just saw your message
I would like to make a model of Alladin from Cartoon
haughty moss
Advanced RVC Inference, by r3gm Huggingface Spaces it does not working with huggingface
odd shale
Advanced RVC Inference, by r3gm Huggingface Spaces it does not working with hug...
Use Ilaria RVC.
odd shale
I would like to make a model of Alladin from Cartoon
Alladin from his movie?
odd shale
yeap
You will must download it's movie with your desired dub.
And isolate it's audio and extract alladin's voicelines
Unless you manage to find it's compiled voicelines on a video
modest delta
ket's say I got 30 second video does that work?
odd shale
ket's say I got 30 second video does that work?
Nope, that will don't work.
Unless you could use OV2 pretrains.
modest delta
gotcha ty ty
odd shale
But 30 seconds of voicelines isn't recommendable.
ripe yoke
Which RVC do I download
I look up RVC download and there’s like a thousand different versions
timber ridge
I need help here accidentally closed out of trhe gradio local host while it was training. Did I mess it up or how can I recover it to do train feature index once its complete
does it even matter still using the same host port
rare gobletBOT
Ayo? @timber ridge level 2 !!!
ripe yoke
Which RVC do I download
I look up RVC download and there’s like a thousand diff...
I need help
It’s like a massive rabbit hole
I keep going to different sites
Links don’t work
Downloading apps
WHAT DO I DO
odd shale
I keep going to different sites
Links don’t work
Downloading apps
WHAT DO I DO
Just download Mangio Fork or Applio.
ripe yoke
How do I download them
Github makes no sense
timber ridge
just crossed 400 epochs. Is there a way I can stop training. Theres no stop option in Gradio
okAY USING some reasoning. using
INFO:AItrioc:Saving model and optimizer state at epoch 400 to ./logs\AItrioc\G_2333333.pth
I found the file that its in at the 400 save mark can I just take this file and copy it. and use this version of the model thats at 400?
odd shale
How do I download them
Github makes no sense
Just press this link if you want to do inference only on RVC: https://huggingface.co/MangioRVC/Mangio-RVC-Huggingface/blob/main/Mangio-RVC-v23.7.0_INFER.7z
brittle wing
Umm.. It looks fine for now but you should check g/mel and loss/d/total
Is this good?
bleak radish
help rvc
rare gobletBOT
Ayo? @bleak radish level 4 !!!
brittle wing
Not exactly a tech question, but is there any way to sort voice models by how popular or well made they are? Like I'm just trying to find good quality male/female models, and I don't really care about the character, just how realistic it sounds
severe sequoia
anyone please what should i do
brittle wing
Can you show me the g/total?
odd shale
Not exactly a tech question, but is there any way to sort voice models by how po...
I think there's no way to sort models by how popular they are. But i mean, i got a girl model you could test xd
odd shale
If D total is going up, then i think it's overtraining.
brittle wing
I think there's no way to sort models by how popular they are. But i mean, i got...
Alrighty, I mean if you wouldn't mind sending it on over I'll give it a test :D
bleak radish
help
old pulsar
prime glen
hey im having issues with using pytorch, this is in powershell
>>> import torch
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'torch'``` i was able to verify the install, some sources say i need to run torch first but theres no good info on how to do so
lagging at cs2
Cs2?
Counter Strike 2?
Are you trying to use W Okada while playing?
Which are your PC specs?
odd shale
my rvc is lagging
Maybe your PC is oversaturated with open programs.
Or your PC doesn't meet the minimum requirements.
Depending what you wanna do, if inference or training
hollow tiger
Are you trying to use W Okada while playing?
it started lagging
voice sound like lagging
but i have rtx 4070 ti
he's lagging only in cs2
i sound like a robot
with rattling
odd shale
with rattling
Which is your CPU?
spare shale
( @odd shale on the Russian channel [#🌏│русский] They wrote the same thing)
hollow tiger
i5-13500
odd shale
it's not a problem
Bro, but you're literally trying to run an AAA game while using voice changer.
If you try to do both things at same time of course one of them will lag.
hollow tiger
but lol
it's cs2
and everything was good before update
i sound like a robot
odd shale
and everything was good before update
CS2 update?
hollow tiger
CS2 update?
yep
hollow tiger
everything was very good before this update
odd shale
everything was very good before this update
Sorry, i don't really know how to solve your issue.
hollow tiger
proper shale
thats the virtual audio cable probably, try installing vac by muzychenko (lite version)
hollow tiger
thats the virtual audio cable probably, try installing vac by muzychenko (lite v...
but i can't delete current
rare gobletBOT
Ayo? @hollow tiger level 3 !!!
proper shale
oh
hollow tiger
thats the virtual audio cable probably, try installing vac by muzychenko (lite v...
thank you very much bro
+rep
it helped
+rep
proper shale
thank you very much bro
yw :)
hollow tiger
spare shale
no rep system... sadly
But we have something else
brave garnetBOT
Made with love by Ilaria 💖
half cove
brittle wing
idk if this is the right channel for this, but I use "AICoverGen", but I have a slight annoyance with it. I'm wondering if I'm just doing something wrong. So basically, sometimes it won't make all of the vocals in an original song into the AI Voice, it would like leave it as the original. It's hard to explain.
I think it's the "Backup Vocals" its not doing anything for
quartz aurora
what is the best rvc to use atm?
spare shale
I think it's the "Backup Vocals" its not doing anything for
For me personally, these are the best I have
brittle wing
For me personally, these are the best I have
wdym
I just checked it completely ignores the Backup Vocals
spare shale
wdym
All My Cover Songs have Back Vocal, which you can’t help but turn off (Because it would be very stupid)
I mean without Back Vocal it wouldn't be a good song
brittle wing
The AI completely ignores the backup vocals and doesn't make it sound like the AI Voice Model
spare shale
The AI completely ignores the backup vocals and doesn't make it sound like the A...
Yes it doesn't change Back Vocal It only changes Main Vocal
proper shale
The AI completely ignores the backup vocals and doesn't make it sound like the A...
yeah because... backing vocals are pain sometimes
brittle wing
But sometimes it doesnt sound right because it just fades into the original voice
proper shale
so it just changes the leads
well then you'd have to filter them yourself
and convert with like, ilaria rvc
brittle wing
Im dumb, but would it not make sense for the AI to Generate the AI Voice for the Backup Vocals too? Im genuinely so confused why it wouldnt do that
dawn nebula
Im dumb, but would it not make sense for the AI to Generate the AI Voice for the...
you can
but it will take some time
brittle wing
How would I do that?
umbral marten
what to do if I put my voice but it doesn’t look like him, this is with all the voices what to do
brittle wing
how do i figure out how many epochs one of my old models is
umbral marten
what to do if I put my voice but it doesn’t look like him, this is with all the voices what to do
bleak smelt
does Easy GUI (for RVC v2, with crepe) (with improved downloader) still work? it doesnt seem to be loading my models anymore
proper shale
Im dumb, but would it not make sense for the AI to Generate the AI Voice for the...
because backing vocals can suck sometimes
to convert
finite galleon
Im dumb, but would it not make sense for the AI to Generate the AI Voice for the...
Rvc only takes single speaker audios overlapping audios like backing vocals or harmonies mess with the pitches and features so the result is praticly shit
proper shale
multiple voices dont act well
umbral marten
what to do if I put my voice but it doesn’t look like him, this is with all the voices what to do
proper shale
does Easy GUI (for RVC v2, with crepe) (with improved downloader) still work? it...
nope, doesnt work
has not been working for months at this point
umbral marten
what to do if I put my voice but it doesn’t look like him, this is with all the voices what to do
bleak smelt
nope, doesnt work
Is there something similar I can use on collab?
rare gobletBOT
Ayo? @umbral marten level 1 !!!
umbral marten
what to do if I put my voice but it doesn’t look like him, this is with all the voices what to do
proper shale
what to do if I put my voice but it doesn’t look like him, this is with all the ...
Bro we don't know
Like... alter the pitch settings?
Idk
umbral marten
bleak smelt
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8Hr?usp=sh...
Does it work with rvc2?
rare gobletBOT
Ayo? @bleak smelt level 1 !!!
umbral marten
this my settings
brittle wing
Rvc only takes single speaker audios overlapping audios like backing vocals or h...
I know but with the song I've used on AICoverGen it seems to be good at separating the backing vocals so I dont understand why it cant just seperately generate the ai voice for both of them
proper shale
Does it work with rvc2?
yea
proper shale
rvmpe as f0 det, increase chunks, decrease extra, mess with tune and use client audio
finite galleon
I know but with the song I've used on AICoverGen it seems to be good at separati...
Now that you need to ask the creator to implement
Wait is sociallyineptweeb not on this server?
brittle wing
Now that you need to ask the creator to implement
I guess, but why that isn't already in by default annoys me 
proper shale
Wait is sociallyineptweeb not on this server?
not sure, prolly not
brittle wing
I guess, but why that isn't already in by default annoys me <a:panic:10804269914...
Cuz it has everything to do that. It seperates the instrumental, vocals, and backing vocals, why then only apply the AI Generated Voice to the Vocals, and not the Backing Vocals too?
proper shale
I guess, but why that isn't already in by default annoys me <a:panic:10804269914...
Eh... it's... understandable
But yeah ig AICoverGen is just for quick & dirty covers, you can use anything else to filter and convert, and then have the option to mix stuff up, which is like a 1000x better
proper shale
Cuz it has everything to do that. It seperates the instrumental, vocals, and bac...
Some songs have harmonies tho... it's complicated
brittle wing
Okay tbh, i've never asked, but which is the most suggested one to use, because I kinda suck at finding stuff myself
I do AI Covers alot
I moved away from running RVC locally because it sounds so shit when I do it on my PC
umbral marten
rvmpe as f0 det, increase chunks, decrease extra, mess with tune and use client ...
its didn't work
brittle wing
I mean I've got a pretty decent computer, but it just doesnt work
brittle wing
Eh... it's... understandable
But yeah ig AICoverGen is just for quick & dirty c...
Which version of RVC is best to use if AICoverGen is just for quick and dirty covers
rare gobletBOT
Ayo? @brittle wing level 3 !!!
proper shale
its didn't work
honestly idk then
proper shale
Which version of RVC is best to use if AICoverGen is just for quick and dirty co...
i think any fork like applio or ilaria rvc works for those
just filter out the vocals ig
molten pecan
ila rvc
https://colab.research.google.com/drive/16LkwvFZeudTpUOsE_6bMjOq2qkxFo8...
Is it back now?
proper shale
Is it back now?
wdym
brittle wing
Is Ilaria a really good one?
proper shale
Is Ilaria a really good one?
i mean... it works?
molten pecan
Is Ilaria a really good one?
The best one I know
proper shale
like applio
molten pecan
And the quickest
proper shale
And the quickest
same speed as applio prolly
brittle wing
And the quickest
Yeah AICoverGen takes fuckin' forever to load.
Also do you have to have a good PC to train AI Voice Models locally? because I tried that before with my own voice and the voice was very robotic and just sounded like me with the oldest microphone in existence but even worse.
finite galleon
Yeah AICoverGen takes fuckin' forever to load.
How long lol
proper shale
How long lol
100-500 secs lol
proper shale
Also do you have to have a good PC to train AI Voice Models locally? because I t...
i have a 2012 MacBook Pro
finite galleon
On installation?
proper shale
just saying
brittle wing
i have a 2012 MacBook Pro
I have a Lenovo Legion with an RTX 3060 graphics card
I got it last year
proper shale
I have a Lenovo Legion with an RTX 3060 graphics card
Wait that's actually very decent
The problem with the model was the dataset then 💀
brittle wing
Yep, recently upgraded from 16 gbs to 32 gbs because I run alot of programs
brittle wing
The problem with the model was the dataset then 💀
probably. But the same dataset worked completely fine training it on google colab
proper shale
huh
half cove
Is it back now?
just add numpy==1.23.5
brittle wing
It annoyed the hell out of me, because back when I was trying to train it, there was obviously problems with google colab and rvc (i heard..?)
molten pecan
just add numpy==1.23.5
Huh?
half cove
Huh?
just use this until hina updates it https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Ilaria_RVC.ipynb
proper shale
I just keep getting an error when trying to inf the audio
maybe didn't select something properly or somethin? hm
proper shale
just use this until hina updates it <https://colab.research.google.com/github/po...
huh
what even happened
half cove
things 
bleak smelt
maybe didn't select something properly or somethin? hm
Don't what that could have been
proper shale
just use this until hina updates it <https://colab.research.google.com/github/po...
try this then @bleak smelt
half cove
i think numpy updated so np.int went poop
proper shale
prolly a ila rvc issue rn idk
molten pecan
@trail iris This person wants to know how to connect SynthV to RVC @pallid hare
Good luck with dat
half cove
also other colabs might have the same problem
just add !pip install numpy==1.23.5 in the download requirements
bleak smelt
try this then <@595767548671295520>
Isn't this just the same?
half cove
like the ai covergen ai one
half cove
Isn't this just the same?
it has one thing changed
desert niche
I think this is broken
https://colab.research.google.com/drive/1NA2GuJ2y-zRfoG3NearNiMCQa8NbidSh?usp=sharing
rare gobletBOT
Ayo? @desert niche level 7 !!!
desert niche
I can't use it
(it's the audio dataset maker thing)
anyone got an alternative or smth I can use
proper shale
honestly there isnt anything quite like this rn
the other one broke down months ago too
half cove
i have one but its shitty 
desert niche
honestly there isnt anything quite like this rn
I don't get how ppl can get hours of isolated dataset if it originally had music in the back
proper shale
I don't get how ppl can get hours of isolated dataset if it originally had music...
just... filtering
demucs dnr, bandit plus
desert niche
just... filtering
I only have fl studio and they added a isolation feature but that's gonna take a while depending on the dataset
well I have audacity too
that's it
proper shale
https://www.mvsep.com moment
create an acc btw ;)
you get bandit plus and other useful models for separation
desert niche
you get bandit plus and other useful models for separation
bandit plus you recommend?
proper shale
bandit plus you recommend?
seems to be good for background noise n stuff
desert niche
seems to be good for background noise n stuff
thanks for the heads up
I appreciate it
proper shale
yw :)
azure marshBOT
RVC Guides (How to Make AI Cover)
brave garnetBOT
⠀
Settings for AMD GPUs 
Don't forget that your models needs to be converted in ONNX!
F0 Det.: rmvpe_onnx (suggested for all series)
7xxx XT cards: 112-128 chunk | +16384 extra
6xxx XT cards: 128-192 chunk | +16384 extra
5xxx XT cards: 192-256 chunk | +8192 extra
RX 580: 192-256 chunk | +8192 extra
RX 570: 192-256 chunk | +8192 extra
RX 560: 256-384 chunk | +8192 extra
Advanced Settings
Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low
⠀
Made with love by Ilaria 💖
tribal widget
Hi i Need an help, what is the best method to avoid disconnection on rvc colab during model training? I can't search well for an help
RVC V2 disconnected (Taken from the guide)
trail iris
<@536370896315613214> This person wants to know how to connect SynthV to RVC <@2...
It's against terms of service.
Just use teto.
molten pecan
It's against terms of service.
Just use teto.
oh wait but, how did you do it?
trail iris
oh wait but, how did you do it?
picked it up off the shelf
molten pecan
picked it up off the shelf
i'm talking about connecting it to RVC in realtime
you told me you did something like that
connecting teto or synthv, idr what
tacit grail
Is there a better MDK-NET Model for Ultimate Voice Remover than the base model?
tacit grail
where do i download that?
oh nvm, i figured it out
any other recommended settings?
viral birch
Anyone can help me?
I tried to download the rvc rose quartz (maria rossana) and i get a "entry not found"
forest sluice
I tried to download the rvc rose quartz (maria rossana) and i get a "entry not f...
what are you using? local or colab?
viral birch
I tried to download the rvc rose quartz (maria rossana) and i get a "entry not f...
Lina*
rare gobletBOT
Ayo? @viral birch level 1 !!!
viral birch
what are you using? local or colab?
Idk
jaunty idol
what does epoch mean
pallid hare
It's against terms of service.
Just use teto.
I was looking forward to use some RVC models with SynthV
Plus what is Teto
I don't think you're talking about the SynthV singer right?
brittle wing
guys I'm getting this on the colab:
tacit grail
with "ultimate voice remover" is 4000 segment size always what you should go with for best results? Im not even sure what that means.
proud elbow
with "ultimate voice remover" is 4000 segment size always what you should go wit...
computation resource load, that's far beyond recommended value even for RTX 4090
pallid hare
oh it is the singer 
How is it against the TOS
You mean SynthV TOS right?
Because RVC is Open-Source
And why so
dawn nebula
with "ultimate voice remover" is 4000 segment size always what you should go wit...
think so, the higher the segment size the better the results
so
tacit grail
computation resource load, that's far beyond recommended value even for RTX 4090
whats the recommended value?
dawn nebula
whats the recommended value?
256
proud elbow
whats the recommended value?
between 256 and 1024
tacit grail
Alright, thanks. Any other recommended settings to get the best results?
proud elbow
1024 for MDX23C would be similar to MDX23C in MVSEP
sage solstice
Hello to you all, im new at that, So, I'm making my first model. And I wanted to do the best I could by putting it on 1000 epochs. But its TOO MUCH!!! And im on Mobile!!! I want to reduce it to 300 epochs precisely because of the long waiting time, can anyone tell me if I can do it while its working or i have to do all the process again? Also, i want know, if i train just this epocs, my model is gonna saying less word? (it has 1h of audio on dataset) Please any angel to save me? 
pallid hare
How is it against the TOS
@trail iris Genuinely curious, by the way
trail iris
You mean SynthV TOS right?
Yes, this is professional audio software that uses paid voice actors who did not consent to using their voices to train or be used as inference in other software that's not explicitly defined in the Contract.
Rvc is open source research software but you must get permission from the voice actor you intend to use. You may train your own voice to use as you wish.
Voices can be copyright
pallid hare
I mean, but what's wrong with using SynthV's piano roll with RVC models? That's basically what I want to accomplish
trail iris
You're still using the acapella as inference with another ai program
Teto is a very specific use case with its own rules.
pallid hare
But what I'm trying to say is
I want a way to run RVC models through SynthV, be it copyrighted or not
Let's say I want to use my own voice and I have a RVC model of it
Or that I have some other RVC model that does not have copyright issues
Like, how do I accomplish that?
trail iris
No, you will want to use diffsinger for that purpose. You can train your own models with it
Maghni ai is also a legal ai voice software in development you can make and sell your own voices. The program itself will be free and you purchase a contract with the singer.
pallid hare
No, you will want to use diffsinger for that purpose. You can train your own mod...
Now the thing is
How do I train a DiffSinger model, and how good your hardware has to be to make something decent?
rare gobletBOT
Ayo? @pallid hare level 3 !!!
pallid hare
Or, if there's a way to run RVC on it (Which I don't think it's possible)... How?
trail iris
Or, if there's a way to run RVC on it (Which I don't think it's possible)... How...
You can use the same training data. They have a discord
forest sluice
HOW TO FIX ILARIA RVC:
- Open the Files section (the folder icon)
- Open the
Ilaria-RVCfolder - Double click on the
vc_infer_pipeline.py, a text editor will open on the right hand side - Scroll it down to line 361 (the number right next to the text editor)
- Replace
np.intwithint
rare gobletBOT
Ayo? @forest sluice level 30 !!!
elfin bay
Ermm guyzs what is this
brittle wing
what alternatives are there for voice cloning songs other than voicemyai.com
azure marshBOT
Suggestions for @brittle wing
forest sluice
Ermm guyzs what is this
does your machine have an NVIDIA GPU?
civic onyx
"module 'numpy' has no attribute 'int'.\nnp.int was a deprecated alias for the builtin int. To avoid this error in existing code, use int by itself. Doing this will not modify any behavior and is safe. When replacing np.int, you may wish to use e.g. np.int64 or np.int32 to specify the precision. If you wish to review your current use, check the release note link for additional information.\nThe aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:\n https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations"
i get this error trying to make a cover in AiCoverGen
i think hina fucked up by accident or downloaded a bad update
because it was working just fine 2 days ago
and now we watch my problem not get awnsered
or not nvm
W staff sorry for doubting you
forest sluice
"module 'numpy' has no attribute 'int'.\n`np.int` was a deprecated alias for the...
Open the Files (folder icon) section in Colab and send me a screenshot
forest sluice
open the Hina_RVC folder, then go to the src folder, and double click on the vc_infer_pipeline.py
civic onyx
open the Hina_RVC folder, then go to the src folder, and double click on the vc_...
aight did that
forest sluice
scroll down to line 368, replace np.int with int
Finally, terminate the running cell (if there are any), then re-run it again
(you have to do this whenever you boot up the Colab session)
civic onyx
(you have to do this whenever you boot up the Colab session)
gonna copy and paste this into a textfile so i can just copy and paste the whole thing instead of having to scroll down
frosty badger
Help, don't let me speak with any voice other than Beatrice
civic onyx
also since you're already helping me i have this problem with this ai voice model that gives me "1 -1 index out of bounds" or something like that
forest sluice
have you tried out other models?
civic onyx
have you tried out other models?
yeah they work but just this one specific model just does not want to work giving me that error message
also terminating would mean using the "disconnect and delete" option in the runtime tab right
get this error everytime i try to use the model "Error
index -1 is out of bounds for axis 0 with size 0:
thanks for the colab fix btw
elfin bay
does your machine have an NVIDIA GPU?
What driver do I dowload
forest sluice
also terminating would mean using the "disconnect and delete" option in the runt...
you can just click on the spinning play button on the running cell
quick jungleBOT
you can just click on the spinning play button on the running cell
tacit grail
is the more epochs mean its just more objectively better? whats the recommended value of epochs?
forest sluice
is the more epochs mean its just more objectively better? whats the recommended ...
depends
You may want to check out https://aihubdocs.github.io/en/rvc-resources/epochs-overtraining--tensorboard/
simple mango
Yo Ilaria keeps erroring out anyclue why that's happening
azure marshBOT
RVC Guides (How to Make AI Cover)
deep bear
rare gobletBOT
Ayo? @deep bear level 2 !!!
deep bear
whats this
north torrent
yo, dont mean to be that guy, but how would you remove a radio-like filter from a voice
brave garnetBOT
⠀
Settings for Nvidia GPUs 
F0 Det.: rmvpe (suggested for all series)
RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra
Advanced Settings
Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low
⠀
Made with love by Ilaria 💖
subtle wharf
Okada RVC -
Difference between with Client, or Server audio?
Which one is better? If server what is S.R, and it's value?
carmine hearth
subtle wharf
Thanks!
brittle wing
halp
lean burrow
C:\Users\Administrator\Downloads\Applio-3.0.6\env\lib\site-packages\torch\autograd_init_.py:251: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed. This is not an error, but may impair performance.
grad.sizes() = [64, 1, 4], strides() = [4, 1, 1]
bucket_view.sizes() = [64, 1, 4], strides() = [4, 4, 1] (Triggered internally at C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\reducer.cpp:334.)
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
After this, no message appears.
nvm it fixed
rare gobletBOT
Ayo? @lean burrow level 2 !!!
carmine hearth
The training speed will be the same for any pre-training model.
트레이닝 속도는 어떤 사전학습모델이든 전부 똑같을거에요
carmine hearth
I would recommend ov2, but I don't know what others think.
If your dataset is longer than 10 minutes, I think you might be better off using the default pre-trained model.
저는 ov2를 추천드리는데, 저 말고 다른 분들 생각은 어떨지 모르겠네요.
데이터셋 길이가 10분을 넘어가시면 기본 사전학습모델을 사용하시는 것도 좋다고 생각해요.
lean burrow
ok thanks :3
brave garnetBOT
⠀
Google Colabs 
⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.
AICoverGen-NoWebUI
Useful for making covers, doesn't inclued 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 💖
lost relic
RVC Quality : low. does it mean quality is low ? why don't just set it high so get better quality
brittle wing
How to train models on AICoverGen-WebUI successfully?
odd shale
RIN E3 VS OV2 What is better?
Only RIN E3 is better if you got a insanely clean audio
odd shale
Look At me!
Look at Me is a banger.
But seriously, on which HF space you got that error?
tame python
But seriously, on which HF space you got that error?
No HF AICOVERGEN HINA MODE
COLAB
tame python
Ok
tame python
Maybe you did something wrong.
I do not know
tame python
Maybe you did something wrong.
finite galleon
Wrong numpy version perhaps
tame python
How can I solve this?
tame python
Wrong numpy version perhaps
How can I solve this?
carmine hearth
Is aicovergen-webui colab still valid? I'm not sure as I only use nowebui colab.
round drift
hey guys i want help
how to use models
like i got john cena model
but how to use it
tame python
Is aicovergen-webui colab still valid? I'm not sure as I only use nowebui colab.
Link?
rare gobletBOT
Ayo? @tame python level 2 !!!
carmine hearth
tame python
https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp=sh...
It’s like this elsewhere too… .
carmine hearth
but how to use it
finite galleon
How can I solve this?
Perhaps stop the running cell, create a new cell, type !pip install numpy==1.19, run the cell, run the web ui cell again
brittle wing
Do anyone know how I can use RBC on Android phone
finite galleon
But again i am not sure if this is a complete fix as i use another colab that works just fine
brave garnetBOT
⠀
Google Colabs
⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.
AICoverGen-NoWebUI
Useful for making covers, doesn't inclued 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 💖
carmine hearth
It's possible to use rvc colab on an android phone, but as far as I know, android phones can't handle rvc.
rare gobletBOT
Ayo? @carmine hearth level 8 !!!
round drift
https://discord.com/channels/1159260121998827560/1203836329172013148/12038363291...
they didnt tell me how to use models they tell me how to make models
or upload
azure marshBOT
RVC Guides (How to Make AI Cover)
molten pecan
It's possible to use rvc colab on an android phone, but as far as I know, androi...
You can use RVC through the cloud
https://aihubdocs.github.io/en/ check Ilaria RVC or AICoverGen NO UI page
For training, use RVC Disconnected
round drift
where can i get a real time voice changer
tame python
It's really annoying... .
brittle wing
i'm using applio V3 and when inferencing it gives me this error
odd shale
i'm using applio V3 and when inferencing it gives me this error
Maybe you should change the name of the file you want to use for inference
brittle wing
ok ill try that
brittle wing
-colab
mossy grail
forest sluice
not dangerous
mossy grail
not dangerous
source?
mossy grail
last time i ignored this i got scammed
sand iris
hang on, what's the message saying ?
forest sluice
This message seems dangerous on the img
mild junco
proper shale
you need a virtual audio cable, check https://rentry.co/VoiceChangerGuide
night swallow
you need a virtual audio cable, check https://rentry.co/VoiceChangerGuide
how do i fix generating sola buffer?
@round fog sorry for ping but i need help
rare gobletBOT
Ayo? @night swallow level 1 !!!
warm needle
Hey guys im sorry but what can I use to transform text-to-speech using my own voice model (downloaded model)
Like voiceovering the text using the downloaded model
rare gobletBOT
Ayo? @warm needle level 3 !!!
half cove
please use these links for aicovergen ui and Ila rvc for the time being
https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Ilaria_RVC_Tempfix.ipynb
https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Hina_Mod_AICoverGen_Tempfix_colab.ipynb
also if any other colabs have a error similar to "AttributeError: module 'numpy' has no attribute 'int'." you can add !pip install numpy==1.23.5 in the requirements to fix it (if you dont know how ping me)
tribal widget
Hi i Need an help, what is the best method to avoid disconnection on rvc colab d...
someone can answer me
half cove
someone can answer me
ngl the best way is just to stay on the tab
just check up on it every few min
cuz it uses captcha now
muted merlin
"module 'numpy' has no attribute 'int'.\nnp.int was a deprecated alias for the builtin int. To avoid this error in existing code, use int by itself. Doing this will not modify any behavior and is safe. When replacing np.int, you may wish to use e.g. np.int64 or np.int32 to specify the precision. If you wish to review your current use, check the release note link for additional information.\nThe aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:\n https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations"
brittle wing
I gotta move it to the c drive?
Not an option
Ain't got a lot of space on the C drive
The previous version of applio worked on the D drive just fine
I'll try that later yea
half cove
"module 'numpy' has no attribute 'int'.\n`np.int` was a deprecated alias for the...
what colab are you using
full tundra
what is this?
half cove
what is this?
what u using?
full tundra
AICoverGen WebUI (Modded by Hina)
half cove
AICoverGen WebUI (Modded by Hina)
look at pins
azure marshBOT
RVC Guides (How to Make AI Cover)
odd shale
AICoverGen WebUI (Modded by Hina)
I suggest you to use the fixed Ilaria RVC colab in pins.
buoyant scarab
# please use these links for aicovergen ui and Ila rvc for the time being
<http...
error in the code of the first link?
during install requirements i get error:
File "<ipython-input-2-f768e9436ade>", line 1
on install requirements do
^
SyntaxError: invalid syntax
buoyant scarab
also i wonder, how does it happen for collabs to be working and then suddenly not working
😭
are unrevised changes in them made or?
half cove
gime 1 sec
buoyant scarab
yea sure
half cove
yea sure
wait a bit for it to update
or just run !pip install numpy==1.23.5 in another cell
molten pecan
also i wonder, how does it happen for collabs to be working and then suddenly no...
Seems like a piece of their software updates without their control, so it messes things up
Or at least that's what happened this time
It's an easy fix though 👍
half cove
or just run !pip install numpy==1.23.5 in another cell
and remove it from the one with error
molten pecan
Also Poppy is a really great artist
half cove
It's an easy fix though 👍
should i fix the error instead of just using a old version of numpy
half cove
Also Poppy is a really great artist
wat
molten pecan
should i fix the error instead of just using a old version of numpy
Yeah I would do that if I was you
half cove
Yeah I would do that if I was you
just need to replace a bit
should i do ila rvc first or
aicovergen
molten pecan
should i do ila rvc first or
Ila
buoyant scarab
or just run !pip install numpy==1.23.5 in another cell
sure
buoyant scarab
Also Poppy is a really great artist
spill
regal aspen
@molten pecan Yo
buoyant scarab
it should have updated by now
didnt fix the issue
rare gobletBOT
Ayo? @buoyant scarab level 1 !!!
regal aspen
So basically I'm wanting to make my first AI model
buoyant scarab
molten pecan
Yea
buoyant scarab
i ran pip install
regal aspen
I'm following a tutorial but the google drive allows you to do all the steps in there instead of opening up on a new tab
buoyant scarab
i ran pip install
(b4 running install requirements)
molten pecan
I'm following a tutorial but the google drive allows you to do all the steps in ...
Can you send a scs of the GD?
half cove
i said to del the one in install requirements
regal aspen
it's 4 steps on how to make the model
half cove
chekc now cuz it should be updated
regal aspen
molten pecan
Hm okay I see
I'm confused about something
What did you download and extract? A model?
regal aspen
What did you download and extract? A model?
that's the thing am tryna make one
molten pecan
that's the thing am tryna make one
Did you download RVC + set up EasyGUI Colab?
regal aspen
doxxed RAHHH 
buoyant scarab
😭
molten pecan
Here's the thing, downloading RVC locally is for using it locally
half cove
put in google drive
molten pecan
Which means using your hardware
regal aspen
me big noob
buoyant scarab
i said to del the one in install requirements
am i blind or is there no pop install numpy
(to be deleted)
molten pecan
And the EasyGUI is for using it through the cloud, if you don't have good specs
half cove
am i blind or is there no pop install numpy
this updated one?
buoyant scarab
this updated one?
yea
molten pecan
me big noob
Np, we all start somewhere 👍
half cove
buoyant scarab
yeah
half cove
fuck
buoyant scarab
same one
buoyant scarab
just to be clear, im talking abt this link:
https://colab.research.google.com/github/poiqazwsx/RVC-Notebooks/blob/main/Hina_Mod_AICoverGen_Tempfix_colab.ipynb
buoyant scarab
gime another sec
yea no worries
molten pecan
then copy the path yea?
Tbh I don't use EasyGUI colab so idk.
Try using RVC Disconnected
https://aihubdocs.github.io/en/ search the guide here
half cove
yea no worries
delete the thing that says on install requirements do
buoyant scarab
delete the thing that says on install requirements do
so far it worked
half cove
so far it worked
tell me if it works
glass cloud
How do I stop training my model ?
buoyant scarab
tell me if it works
it lanuched the GUI, but it doesnt find my .pth model (?)
buoyant scarab
half cove
and what are you using
buoyant scarab
did you download it?
i imported my pth directly into the directory
buoyant scarab
and what are you using
google colab?
half cove
i imported my pth directly into the directory
into what folder
half cove
google colab?
nah dw
half cove
did it work before?
buoyant scarab
i did that before and it always worked
half cove
buoyant scarab
ill try to use upload model tab in gui
half cove
i did that before and it always worked
try importing it using huggingface
glass cloud
How do I stop training, I just close the cmd?
buoyant scarab
ill try to use upload model tab in gui
it gave me an error but ultimately worked lmfao
glass cloud
Does model training last forever or is it up to us to stop it?
rare gobletBOT
Ayo? @glass cloud level 2 !!!
buoyant scarab
Does model training last forever or is it up to us to stop it?
rare gobletBOT
Ayo? @buoyant scarab level 2 !!!
half cove
it gave me an error but ultimately worked lmfao
what error?
buoyant scarab
what error?
directory already exists or whatever
half cove
buoyant scarab
it always gave me that error which was annoying so i switched to manually dropping pth to the folder
molten pecan
How do I stop training, I just close the cmd?
Yes, close the cmd
hot gorge
Hello, I have a question: how to use AI voice to text to speech?
split bear
if i'm training a model with a target sample rate of 48k, should i convert my input data to 48k too? currently it's all 44.1k
odd shale
Hello, I have a question: how to use AI voice to text to speech?
You can use Ilaria RVC, it has TTS added.
odd shale
if i'm training a model with a target sample rate of 48k, should i convert my in...
Nope, it will won't
It will only add innecessary noise to your final model result
I suggest you to check your dataset's audio spectrogram in Spek and check it's true sample rate
If it's 16khz, then 32k
If it's 20khz, then 40k
If 24Khz, then 48k
split bear
thanks didn't know that spek existed
rare gobletBOT
Ayo? @split bear level 1 !!!
split bear
as far as i can tell it says it's alr 48k?
half cove
send image
rare gobletBOT
Ayo? @half cove level 144 !!!
split bear
half cove
yep
split bear
then why did audacity export as 44.1k by default... odd
guess i'll reexport as 48k then
limpid silo
hey guys im unzipping VoiceConversionWebUI thing from the guide and tons of files cant be extracted cause they are unsupported. is this normal?
or should i be worried about this
molten pecan
hey guys im unzipping VoiceConversionWebUI thing from the guide and tons of file...
Unsupported?
Wdym
@half cove any idea?
limpid silo
wait can i not upload image in this channel
half cove
u using 7zip?
limpid silo
just one thats built in on windows
half cove
use 7zip
limpid silo
it says "the entry [file name] cannot be extracted from archive because it is an unsupported file
rare gobletBOT
Ayo? @limpid silo level 1 !!!
limpid silo
alr thx
sand iris
Or you can download winrar xd
not sure that even works
odd shale
not sure that even works
It works well for me.
half cove
It works well for me.
messes up for some people
woven bear
Hello!, does anyone know how I can use RVC on Android? Before, I used to use it through code and connecting my drive account, very easy. Just to download audios, can you please help?
split bear
so i just crashed my computer trying to process a dataset where i'd combined all the files into a single large one... can i split it up anywhere, or should i make sure to split during pauses (between words)?
lilac pilot
how come i keep getting glitchy sounds when processing some acapellas with UVR. when i run dereverb some files sound perfect and the other are a glitched out
molten pecan
Hello!, does anyone know how I can use RVC on Android? Before, I used to use it ...
https://aihubdocs.github.io/es/ mira la página de Ilaria RVC 👍
molten pecan
so i just crashed my computer trying to process a dataset where i'd combined all...
The process of cleaning datasets can be automatized
https://aihubdocs.github.io/en/ check the datasets page. Read the cleaning chapter
Also you don't need to split it, using a single one is fine
molten pecan
how come i keep getting glitchy sounds when processing some acapellas with UVR. ...
That's artifacting, and it depends on several factors
molten pecan
https://aihubdocs.github.io/en/ check the datasets page. Read the cleaning chapt...
Check the artifacting page to learn more
And how to fix
eager siren
it is not creating the audio with the model's voice
split bear
https://aihubdocs.github.io/en/ check the datasets page. Read the cleaning chapt...
ooo, thanks for the guide I missed that section
I wasn’t able to keep it all as one file tho, it filled my memory and crashed the computer lol (during preprocessing)
molten pecan
it is not creating the audio with the model's voice
Double check if the audio is in said file path
Ensure the path doesn't contain spaces/special characters
molten pecan
I wasn’t able to keep it all as one file tho, it filled my memory and crashed th...
Good luck
split bear
will it harm the dataset quality if i split the file in the middle of words?
there’s over an hour of data lol so that’s prolly why it crashed it
molten pecan
will it harm the dataset quality if i split the file in the middle of words?
In my experience yeah, it does
Also, why that long though?
20 mins is perf, as long as the quality is good
split bear
uhhh i wish there was a way to edit all the files without combining them lol
quality should be fine it’s straight from the game. I just thought more = better
molten pecan
quality should be fine it’s straight from the game. I just thought more = better
Hm not necessarily, only at some extent
Quality matters a lot too
You'll get better results with a HQ 5 min dataset, than with a 20 min poor quality one
Usually iirc past 40 minutes is just unnecessary. It won't bring anything new to the table
split bear
do u think 10-20min files wouldn’t crash the computer?
molten pecan
do u think 10-20min files wouldn’t crash the computer?
I mean likely yeah, since it won't require that many resources
@brittle wing @honest pond terminen de escribir
https://cdn.discordapp.com/emojis/1106933951383228466.webp?size=64
azure marshBOT
RVC Guides (How to Make AI Cover)
honest pond
it is not creating the audio with the model's voice
the name needs to be short and whit no spaces
odd shale
the name needs to be short and whit no spaces
Y a tí porque te demotearon?
Porque exactamente?
Ya no quiso seguir siendo Helper?
tame python
https://youtu.be/V7E7Y_iVixQ?si=-8ceAr4R9Gqb1bI1
It comes out here line distribution layout tutorials please…..
rare gobletBOT
Ayo? @tame python level 3 !!!
spiral hamlet
Someone have a gradio/collab link please??
rare gobletBOT
Ayo? @spiral hamlet level 1 !!!
blazing ibex
Error
"module 'numpy' has no attribute 'int'.\nnp.int was a deprecated alias for the builtin int. To avoid this error in existing code, use int by itself. Doing this will not modify any behavior and is safe. When replacing np.int, you may wish to use e.g. np.int64 or np.int32 to specify the precision. If you wish to review your current use, check the release note link for additional information.\nThe aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:\n https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations"
How to fix this? 😦
hidden drum
Porque exactamente?
for ai remaster how do you extract a clean acapella to use to make an ai remaster bc im trying to make an ai remaster of juice wrld rolling loud @odd shale
odd shale
for ai remaster how do you extract a clean acapella to use to make an ai remaste...
AI remaster?
hidden drum
AI remaster?
yeah
fluid matrix
why did you tag me and then delete and tag leo 💀
hidden drum
why did you tag me and then delete and tag leo 💀
mb
hidden drum
why did you tag me and then delete and tag leo 💀
didnt know if you would of responded
odd shale
I suggest you to record vocals yourself.
fluid matrix
you deleted it instantly, you have to give me some time to respond 😭
you are trying to make a ai remaster, what would you even need acapella for?
hidden drum
you deleted it instantly, you have to give me some time to respond 😭
mb i wont delete msg next time lmfao
hidden drum
you are trying to make a ai remaster, what would you even need acapella for?
yeah
but the problem is im not black
fluid matrix
the whole point is to make the vocals yourself
hidden drum
so i cant say the word if juice says the n word 😭
fluid matrix
then just censor it ig
hidden drum
you are trying to make a ai remaster, what would you even need acapella for?
for rolling loud of juice thats the acapella i need
fluid matrix
bro you cant just use lq acapella and run it through ai thats not how it works
the whole point of ai remasters is to make the vocals yourself and run that through the ai, then make the beat and mix/master
hidden drum
bro you cant just use lq acapella and run it through ai thats not how it works
abt like uvr can i use that
odd shale
bro you cant just use lq acapella and run it through ai thats not how it works
That's what i tried to tell him.
hidden drum
then just censor it ig
censoring it tho will be so bad tho it will ruin the whole flow
😭
odd shale
then just censor it ig
Bruh, he can say the word, he'll anyway will pass his reference vocal over a Juice model.
fluid matrix
yeah
hidden drum
well either you say it or you dont 💀
i aint black im asian asain cant say racist stuff tho
odd shale
It's not like he wanted to make an offensive song with a voice model of Jimin or Jungkook
odd shale
i aint black im asian asain cant say racist stuff tho
Srsly? Bruh, if you just record a vocal and say that word it shouldn't be a problem.
I mean, it's kinda pointless to say that knowing you probably do that in your home
hidden drum
can you send a vid of what ur trying to remaster?
can i send in dms its copyrighted as hell
fluid matrix
you can send a youtube link
odd shale
But ehh, we don't care if you say that word or not.
odd shale
i aint black im asian asain cant say racist stuff tho
Altho is kinda idiotic to censor words like that in a Juice WRLD song, it would ruin the vibe.
hidden drum
https://www.youtube.com/watch?v=DhfGsBjUrjw&t=0s
im tryna make my own one tho
so yeah
fluid matrix
then just do it 😭
hidden drum
Don't repeat it, just do it.
K
tame python
Please tell me how to add a download button in Gradio😭😭😭Please😭
molten pecan
Please tell me how to add a download button in Gradio😭😭😭Please😭
Are you creating a Gradio layout?
Or are you trying to download a model?
tame python
Are you creating a Gradio layout?
I want to put a download button here
molten pecan
I want to put a download button here
Did you try tapping the three dots?
Or try going on desktop mode
tame python
Did you try tapping the three dots?
I tapped it and the speed settings appear.
tame python
I’m sorry
molten pecan
I don't know the answer. Maybe screen record and extract the audio
Or use another web browser
tame python
I'm using an iPhone and it's very inconvenient that EASYGUI Colab doesn't have a download button.
Here
…..
azure marshBOT
Suggestions for @tame python
rare gobletBOT
Ayo? @tame python level 4 !!!
tame python
Ok
untold steeple
I could use some help with my mac Applio
Using Mac m1 running 14.2.
Launched Applio loaded a few different models I got on here, by dragging .pth and .index files. Applio copies them and lets me load them. When I load my voice and convert, it looks successful. I get an “output” audio file bottom right, but it sounds the same as source file, no changes,
No matter when model I use, get the same error in my terminal:
Traceback (most recent call last):
File "/Users/Documents/Applio-main/rvc/infer/infer.py", line 6, in <module>
from vc_infer_pipeline import VC
File "/Users/Documents/Applio-main/rvc/infer/vc_infer_pipeline.py", line 7, in <module>
import pyworld, os, faiss, librosa, torchcrepe
ModuleNotFoundError: No module named 'faiss'
peak hare
can someone send me the link for amd download 64bit
woven bear
https://aihubdocs.github.io/es/ mira la página de Ilaria RVC 👍
Gracias 🫂
peak hare
nvm
forest sluice
go to vc_infer_pipeline.py file in Ilaria RVC folder, look for the line that has: f0_coarse = np.rint(f0_mel).astype(np.int). Replace the np.int in that line with int
deep bear
what folder im using google collab
forest sluice
Alright, in your google colab tab, you see that folder icon on the left hand side, click on it, a file manager will open, there should be an Ilaria-RVC folder there, click on it, then scroll down, you'll see the vc_infer_pipeline.py file, in which you'll double click on it, a text editor will spawn on the right hand side
regal orbit
There are so many versions, which version is best?
deep bear
go to `vc_infer_pipeline.py` file in Ilaria RVC folder, look for the line that h...
this?
tame python
Is there anyone who can write lyrics and sing?
simple mango
Wow a me
but the real reason I'm here is that Ilaria keeps erroring like with no reason ata ll
simple mango
What for mate
simple mango
Kk well usually I make beats and idk what Nmixx is
tame python
Kk well usually I make beats and idk what Nmixx is
Kpop girl group
simple mango
ah
Well I'm like super busy but if you can't find anyone else I'll do it for shits and giggles
rare gobletBOT
Ayo? @simple mango level 3 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
tame python
Well I'm like super busy but if you can't find anyone else I'll do it for shits ...
I'll send you a DM
simple mango
k
split bear
grad.sizes() = [64, 1, 4], strides() = [4, 1, 1]
bucket_view.sizes() = [64, 1, 4], strides() = [4, 4, 1] (Triggered internally at ..\torch\csrc\distributed\c10d\reducer.cpp:334.)```
Is this something I should be concerned about?Nope! that's the usual nonsense
The original RVC makers apparently went with it
forest sluice
yes
spice thorn
Colab
azure marshBOT
wanton patrol
Can someone please tell me how to install this please? https://www.weights.gg/models/clqy0k8jb0046chk5ye5r2q0x
rare gobletBOT
Ayo? @wanton patrol level 1 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
proper shale
Can someone please tell me how to install this please? https://www.weights.gg/mo...
make sure you're signed up, then just click on the three horizontal dots and click on the download button
odd shale
Can someone please tell me how to install this please? https://www.weights.gg/mo...
Press those 3 dots.
split bear
should these sections be removed from the dataset?
rare gobletBOT
Ayo? @split bear level 2 !!!
odd shale
should these sections be removed from the dataset?
Maybe you can remove them if you know how.
split bear
Maybe you can remove them if you know how.
spectral de-noise yeah.. good idea
is it still worth it with such a big dataset? i could also just remove those sections entirely i guess
odd shale
is it still worth it with such a big dataset? i could also just remove those sec...
You can do that too.
It's not necessary to use an giant dataset.
wanton patrol
make sure you're signed up, then just click on the three horizontal dots and cli...
Thanks but i ment how to setup and install sorry for not being specific
regal orbit
Suggestions for <@1204403827718099004>
Please tell me which version is best
brittle wing
I assume I can't use these on mobile
split bear
does this matter? not sure what dithering is, but feels like it could affect the data
odd shale
does this matter? not sure what dithering is, but feels like it could affect the...
On Bit depth select 32-bit
Do not select 16 bit
proper shale
Thanks but i ment how to setup and install sorry for not being specific
this has everything you need to know
molten pecan
@ornate hatch https://aihubdocs.github.io/es/ buscá acá la página de Ilaria RVC
Ahí está la guía de uso + el link actualizado
Es fácil de usar en teléfono también
ornate hatch
Oka, gracias!
molten pecan
Oka, gracias!
De nada! Ante alguna duda me etiquetás 🙏
half cove
does this matter? not sure what dithering is, but feels like it could affect the...
i might be wrong but i think you should only use dither when lowering the bit depth on the audio
so just keep it disabled
split bear
yeah i found the problem: the original data is 16bit
but then when cleaning it somehow it gets changed to 32bitfloat
so now idk how to save it ;-;
wanton patrol
https://aihubdocs.github.io/en ig
Tysm
half cove
but then when cleaning it somehow it gets changed to 32bitfloat
32bit float is fine
i always use 32bit float
split bear
32bit float is fine
even if the dataset is 16?
half cove
even if the dataset is 16?
yep
split bear
ohk, thx
minor vapor
I'm getting this error in AICoverGen_colab.ipynb but it said that I have no Nvidia driver, but I download the newest one...
It not working, when it's making the song with the Ai voice.
molten pecan
Do this
Press the downward arrow on the right corner
And click delete runtime
And reload the page
keen mica
Is the best rvc w okada or are there better alternatives for quality?
minor vapor
Press the downward arrow on the right corner
what arrow
minor vapor
it did not worked
I think I saw the GPU is not backed
what that mean?
oh, ok, but why the code said that I don't have a GPU but I have one.
rare gobletBOT
Ayo? @minor vapor level 2 !!!
minor vapor
what other colab you mean CoverGen_No_UI.ipynb
azure marshBOT
Suggestions for @minor vapor
limpid silo
what is better when you train models? large data with worse quality or small data with good quality?
I am trying to train singer that i know. i have lossless songs from album (around 10 songs) and her entire youtube channel downloaded (70 vids). I am not sure if i should just use songs from album or if i should use youtube too.
rare gobletBOT
Ayo? @limpid silo level 2 !!!
limpid silo
thank you @rare goblet!!
but 10 songs is not enough right? idk what the recommended length is
10 songs should be around 40 min
ah ok
minor vapor
I think I'm going to wait tomorrow, but thanks for the help.
rancid rapids
how come all the models that I'm using from #1175430844685484042 all sound like shit and are really laggy. Surely I'm doing something wrong but idk what
lilac spear
hello. i was wondering if there are any other programs i could use to make an RVC model instead of doing it in RVC itself
azure marshBOT
RVC Guides (How to Make AI Cover)
dawn nebula
hello. i was wondering if there are any other programs i could use to make an RV...
just rvc itself i think
dense sapphire
does someone know why when i run the go-web.bat it opens, and then shows: C:\Users\user1111\Desktop\RVC1006Nvidia>pause
Press any key to continue . . .
and doesnt do anything after that
polar hound
so if im using a custom pretrained model, do i uncheck just the "pretrained"? wish devs would make things a little clearer yk
nvm unchecking pretrained itself just doesnt actually train it at all
thats really silly
prisma kettle
You can use UVR to remove music instead of vocals, right? How do I do that
rare gobletBOT
Ayo? @prisma kettle level 1 !!!
polar hound
You can use UVR to remove music instead of vocals, right? How do I do that
uvr gives u both vocal and instrumental outputs
rare gobletBOT
Ayo? @polar hound level 3 !!!
prisma kettle
Is there a different model I should use rather than the default? I rememmber downloading some a while ago on my other computer
lilac spear
I need some help. I've been trying to vocal train to make a model on my own several times, but never managed. It literally takes me ALMOST HALF AN HOUR to make just ONE epoch. I dont understand what the hell im doing wrong... i got 8Gb of GPU, 11th generation core. Can someone please help me figure out what to do?
Im losing my goddamn mind over this, please.
frozen tendon
I keep trying to convert an audio and rvc keeps crashing. I shortened the audio and it still crashes. I don't understand, it worked fine before this.
rare gobletBOT
Ayo? @frozen tendon level 1 !!!
lilac spear
Nvidia 3050
rare gobletBOT
Ayo? @lilac spear level 2 !!!
lilac spear
It's a gaming pc, i've played Cyberpunk on very high settings on it no problems
jaunty shale
Question, do I have to install Applio only on C drive or can I download it on D drive too?
(had it before with old version and it worked fine, this new version on github it says that I have to put it on C drive)
lilac spear
Please write me on private if you can help me...
jaunty shale
Does it really affect anything if I put it on D drive instead? (it didn't before when I had older version)
lament ermine
whats best 3 rvc GUIs?
jaunty shale
From my knowledge, when I converted audios into AI ones, it was fine even when I had it on D drive
I had no issues at all
lilac spear
I'm not freaking joking look at this
jaunty shale
Noted.
jaunty shale
I'm not freaking joking look at this
assuming you have RTX?
lilac spear
assuming you have RTX?
Yes
jaunty shale
how much vram/video memory you have?
lilac spear
- i tried setting it to 20 but i got the CUDA memory error
jaunty shale
how much vram/video memory you have?
if you don't know, press win + r and type dxdiag, go to your monitor and find your dedicated GPU with info on it. There should be info about the VRAM
rare gobletBOT
Ayo? @jaunty shale level 6 !!!
jaunty shale
Ayo? <@521718007039262760> level **6** !!! <:lfg:1159355870119993496>
shut
lilac spear
how much vram/video memory you have?
7,9 GB of GPU
jaunty shale
7,9 GB of GPU
Well that explains a lot.
sort of.
I'd recommend training in colab
like I do
lilac spear
what do you mean?
prisma kettle
My vocals made in UVR don't sync with the original track when I try to line them up
in adobe premiere
jaunty shale
Basically what you do, is the training algorithm uses your gpu to run it, maybe the ram won't be enough?
I am not sure myself, but if you looking for faster way to train a model, I'd recommend google colab.
lilac spear
Basically what you do, is the training algorithm uses your gpu to run it, maybe ...
i dont know how to do it
jaunty shale
i dont know how to do it
sand iris
8. i tried setting it to 20 but i got the CUDA memory error
You're not gonna get anything higher than 6 on 8GB cards
jaunty shale
You're not gonna get anything higher than 6 on 8GB cards
yeah, basically. Better explained
lilac spear
fuck my life..
prisma kettle
My vocals made in UVR don't sync with the original track when I try to line them...
does anyone know why this might be
lilac spear
hey don't worry, you will get around with the collab just fine
can you please give me a hand on DM? english it's not my first language
jaunty shale
can you please give me a hand on DM? english it's not my first language
Sure thing
limpid silo
hey guys im currently getting samples for model that im about to train
especially around1:25 theres layers and echos which i assume it will make my model worse?
proper shale
especially around1:25 theres layers and echos which i assume it will make my mo...
yeah
also dont upload acapellas here directly next time, thx
limpid silo
ah mb
proper shale
you can use like... uvr's bve
for the lead vocals
and uvr de echo de reverb to get the reverb/echoing out
;)
odd shale
and uvr de echo de reverb to get the reverb/echoing out
De echo de reverb is better than MDX De reverb
novel arch
which is the better pretrain. OV2, RIN_E3 or Original
brittle wing
Can anyone who's familiar with RVC Train colab can you pls tell me where's the link of Huggingface Space on RVC Train because I can't use Gradio there because the Gradio from RVC Train is now for paid users
analog obsidian
which is the better pretrain. OV2, RIN_E3 or Original
use ov2 if you have a small dataset (1 to 5 minutes)
rin e_3 is very sensitive to noise, if you have a 10 minute dataset or more use original tbh
dawn nebula
Can anyone who's familiar with RVC Train colab can you pls tell me where's the l...
is this rvc disconnected or sth else?
azure marshBOT
RVC Guides (How to Make AI Cover)
limpid silo
hey guys
looking into this guide https://aihubdocs.github.io/en/rvc/local/mainline/ rn, and i just have one question
so it says inputting incorrect one might screw u the final quality. so should I change the sample rate of datas using audicity?
proper shale
De echo de reverb is better than MDX De reverb
mdx de reverb better because no 17,5khz limit
but it depends
proper shale
so it says inputting incorrect one might screw u the final quality. so should I ...
not necessarily
proper shale
dw
when u preprocess, it changes the sample rate to that
id use 32k, always
less noise overall, no big quality difference because its all just 768 layers anyway
proper shale
idk mine doesnt even have 32 lol
rvc v1 moment
limpid silo
sorry another question
rare gobletBOT
Ayo? @limpid silo level 3 !!!
limpid silo
my terminal doesnt say end process. is it just processing stuff?
sorry this is command prompt
sharp mason
Hey guys, so I'm trying to make an AI cover but every google drive link I use won't go through. I'm using a google colab and it seems to be like that for every colab version that I have tried. What do you reccomend?
I'll try some more links
odd shale
I'll try some more links
Umm.. Try Ilaria RVC
Any model you want to use that is on a drive link, download it and upload it to your HF account.
(Huggingface)
sharp mason
Huggingface links are still working?
rare gobletBOT
Ayo? @sharp mason level 1 !!!
carmine hearth
sorry this is command prompt
Try renaming your audio files with all numbers.
Spaces, special characters, etc. will prevent the dataset from being recognized correctly.
limpid silo
alr i will try that next time. as of right now i just assumed that it worked and currently training the model
I think it worked i assume its just their (dev) way of doing string manipulation
brittle wing
its not working for me any more any time try to make an ai cover it closes its self
elfin lantern
I know this sounds stupid but when I try to select an audio file, it just won’t let me. I’m on my iPhone since I don’t have a laptop and when I try to, it’s grayed out and won’t let me do anything. It worked on my iPhone back when you put the file into the colab but not anymore
dawn nebula
!howtoask
crystal grotto
Can someone catch me up on why I would be receiving this error using Hina Mod Collab? "Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx"
rare gobletBOT
Ayo? @crystal grotto level 1 !!!
dawn nebula
Can someone catch me up on why I would be receiving this error using Hina Mod Co...
did u enable gpu runtime
crystal grotto
~~I believe so. I ran everything. This appears in gradio in the final step when I click 'generate'~~NVM. The temporary link in the pinned message here works.
I've never had to download any drivers when using rvc collabs. Is this new?
elfin lantern
How do we send screenshots if we do not have image perms?
brittle wing
is this rvc disconnected or sth else?
Something else, the colab is called "RVC Train"
dawn nebula
use rvc disconnected
dawn nebula
How do we send screenshots if we do not have image perms?
you have img perms in this channel when u reach level 1, if im not wrong
forest sluice
Can someone catch me up on why I would be receiving this error using Hina Mod Co...
Click on Edit > Notebook settings?
limpid silo
hello. according to the guide (https://aihubdocs.github.io/en/rvc/local/mainline/), when i do not see stop trainning button, i should just close the cmd window. after that what should I do? should i just proceed to .index and pth file and start inferencing
i see it under log folder. is this something else?
added_ .index
so i can just close the cmd and i should be good to go?
yeah thats the weird part
my weights folder looks same as before i ran the training
just making sure, its root/assets/weights right
brittle wing
why when i try to use voice change it says "Frequent errors occur. Please check if the model of the framework being targeted is loaded
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
my bad