#✨│ai-help
1 messages · Page 17 of 1
How long is the song
Technically it's not a song. ^^;
It's kinda important for my OC
But it is on this %
Oh
Anyone could send me the link for RVC Applio Colab plz?
theres rvc disconnected for free colab users
or ai-cover gen nowebui for inference
-colab @loud stratus
Suggestions for @loud stratus
- AICoverGen-NoWebUI, by Ardha Google Colab
- RVC Disconnected, Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- RVC-HF, by r3gf Huggingface Spaces
- AICoverGen, by r3gf Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
- Advanced RVC Inference, by neuclya Huggingface Spaces
You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.
thx man
altho AICG-NO-WEBUI has disconnection isses
oh fr?
So be advised
yeah, it does
got restricted recently. they shouldve changed more code on it
kk ill keep it in mind
Now it's going well
I think it's much better eventually
Reeee it's going well
is there any way to bypass colab runtime disconnect issue?
@proper shale come on.. ;(((
Ayo? @uneven elk level 3 !!! 
I think the audio might be too long
I think its like
10 min max
AICG is for like
Music
I think u should try like... RVC-HF
It might take like
Hours
Considering how long it is
Oh
inference?
Alr I do RVC-HF
Yeah, sorry :(
yeye
Is there the link for it?
theres also ilaria rvc
Doesn't the hina colab still work?
i did like 2 minutes audio in 800s
DAMN
with RVC-HF
it also depends on ur connection etc
maybe youre more lucky than me
im currently using ilaria's rvc
dunno if theres so much difference tho
I dont think Internet plays a part as all the computing is done over cloud
yh but if u crash during the process
can anyone help me with this?
True
what does this mean
send it
assert cpu_threads>0, "CPU threads not allocated correctly."
do ya have a screen or something?
yeah but i cant send it
Ayo? @halcyon oxide level 1 !!! 
oh thanks
@proper shale is it model interface?
for RVC-HF?
im trying to create an ai voice model
mhhh its a cpu problem
??
wym?
is there anything i can do?
if u did all the steps correctly
also have u already duplicated the space, maxi?
the only thing you can do is disconnect this runtime and retry in a new one
before doing that
before doing that make sure that colab's cpu is connected @halcyon oxide
u have to click on refresh
after downloading the model in Resources
And then refresh when uploading the model
how do i do that?
Select both stuff from dropdown
top right that should be something like this @halcyon oxide
the modell didnt work
if you see this it means that the cpu its connected
wha?!
because of the %
but it says error?
they work
covert process or model download?
look
where?
see this
yh?
thanks for the help
dm me if ya got any problems
now try copying the link from the download button
oh i see the problem
ppl added spaces to files...
k
k
copying link (make sure anyone w/ link can see so applio downloads it correctly)
and doing the model installation stuff
its downloading
ooh nice
@proper shale wha??
ignore those
k
k
42 min is 
fr
its gonna take long af
error.. @proper shale
^
how about u do it for me?
Ayo? @uneven elk level 4 !!! 
because i gtg bed so. ;((
i dont even think i can
just tryy
42 minutes is way too much 😭
;((((
bruh i closed the pc like 10 minutes ago and im tired asf
why tha audios is that long?
to make it turn out well it should be splitted in 3 minutes max 4 minutes lil parts
Ayo? @brittle wing level 1 !!! 
waa
RVC Guides (How to Make AI Cover)
Translation by country
@half cove
I downloaded a voice model that came with two paths and a config, how do I use it?
Are you a Mac user?
I don't know what you mean by paths and a config
two path files and one .json
idk how do i use that
Shit I don't know what a path file is 😭
I only know where to paste the .pth and .index files
yea, thats a config.json and two .pth
Ayo? @undone crow level 1 !!! 
where all am i supposed to submit models btw?
I Heard about FCPE, what is that? new infering method?
Do you guys know how to make a model thru Rvc? mine keeps getting an error (im using the hugging face ver)
what the error?
just error
same
when downloading the model just says "There's been an error."
I'm getting an error as well.
Nope, ive done everything your supposed to in the colab ver, and at the conversions it just errors
oh collab stuff, i though RVC local
im doing it thru huggin face tho-
Its not really a new f0 method
Its just a method thats not implemented to rvc officially.
can someone help me to get the the Ai to work whilst using voicemeter?
Ayo? @bright veldt level 2 !!! 
what is Formant shift inference audio?
How do I make my own model?
How many audio files do I need to train a voice model?
What do I do when my voice model loads infinitely? Or is it normal to take a long time?
how can i find who made a model i have a few questions
Ayo? @green birch level 7 !!! 
Does anyone know how can train ai local with amd?
whats the best rvc fork right now?
FOR ANYONE GETTING 'NoneType' object has no attribute 'tobytes' in Applio
Under the index dropdown, click on your index file again. For some reason this fixes it
I downloaded the rvc 7017 version I believe and I am running it on mac, when I installed all the requirements and started the web gui, it works, but when I click convert on the web gui in the inference page, it shows error, and I look back at terminal. and it says that there is an error in line 14 of file my_utils.py, the line is: ffmpeg.input(name, threads=0) if I remember correclty
Can someone please help me with this?
Also, please dm me, because I will get lost trying to find the reply
is this the point to stop training????
Is it possible to get an index file for an already made model? The colab I was using worked without one before but now I need an index and using one from another model doesn't seem to work
use this easy colab
u must use the index of your model only
If the loss value keeps going down, can I continue learning?
By The Way, can i change a model name and it Index?
i tried this just re-name the pth & index, but i not read the index when do inferring
or we can't re-name it?
rename the model as u want but without dots or under score and put the index in a folder with the same name to automatically find it when choosing the model
is this the point to stop training????
can you send a picture of the loss g/total graph as well
also what happened between 8 and 10k steps
Where can I see my loss g/total? Is it the same as loss/d_g/1?
search up loss/g/total at the top bar in tensorboard
should be the first result there
Is the graph between 8000 steps ~ 10000 steps normal?
?
shouldnt be there really, look at your cmd prompt to right before and right after that point
what does it say there
what is the value of the epoch just before the big dip
If the graph is weird, do I have to start over?
ive honestly never seen that before... id continue training and see how the result is. the rest of the graph looks great
where is the big dip???
I'll try to make my deductions in a few minutes, please be normal.
just before 10k steps
yeah ive never seen it skip like that
i don,t know now because i resumed training after 300 epochs so i losed the past informations
that would make sense, i personally use mangio rvc fork
i meant on tensorboard itself. if you hover over a section of the graph, it gives you a value
like in this photo, the value was 8.902, what was the value of the epoch just before that
in that case, id keep training the model
its not done training at this point
see if it can get the total/g any lower
i use a big data so i think it may needs more than 500 epochs
its not exactly a set value, mainly is about when the graphs in tensorboard start to reach a minimum value
like look at this expertly crafted graph here
the minimum value always comes early as here on 10k 😦
thats probably where id stop training, as its a sound minimum value
where the graph has flattened out
train for a while longer to see if it eventually goes down
guys,need some help,how do i fix “connection errored out”
if it doesn't then the value at 10k there might be the best bet
or something as close in value as you can get to that one
ok thanks alot
is this with online training?
no
local training then?
im using RVC1006AMD_Intel,my hardware is all AMD
Ayo? @ocean kraken level 1 !!! 
local conversion
i tried to use model
hmm not sure then, i thought amd gpus couldnt train
could you try training with non local methods then?
i am not training
yes
if you can get it to work locally then, id try this
Go to Ilaria_Rvc_v2 and create an account on Hugging Face. Duplicate the space if there are queue issues Click on "Download Models" and input the URL of the desired model from Forum AI Voice Models and Model Link Sites like Weights.gg. Go to "Inference" and click on "Refresh" next to "Choose the...
🥲 but they say it support AMD
not sure what the deal is with amd cards
can someone help me with the rvc thing i downloaded it and keeptrying to luanch at it just opens console briefly then does nothing (please dm if you knwo hwo to help)
pls dont ignor me
i think it is a new lowest point
heyyy the VB-AUDIO doens't works it is like on mute someone knows whats happening?
it's rmvpe problem,i can use pm and harvest now,thank you
Ayo? @ocean kraken level 2 !!! 
i got the right terminalk open but not sure where/how to type this in
so resume training????
are you training locally
no online
Ayo? @grand willow level 5 !!! 
gotcha
i got 11
see if you can test the model at these 4 points, see what sounds best
hard to say how it will sound from just a graph, i usually like to use the graph to pick out points id like to test, and see which comes out with the best one
ok i,ll try that but i think it needs good tests because results may be near
what do you mean by results may be near
judging good results of tests may be diificult
not too difficult, ones that are bad will just sound uncanny if that makes sense
or off in some way
if you have trouble yourself picking out the ai from the real voice, its probably good
thanks for your efforts
I trained a model and now i'm trying to convert the voice but it's showing error
Ayo? @untold zodiac level 1 !!! 
may be you shoosed another index
idk what i did wrong
can I dm you the ss of what's happening
Still Relevant?
Which to select
i'm trying to use RVC GUI but i have this error when i try to convert a file:
No supported Nvidia cards found, using CPU for inference
Any idea of the issue ?
Ayo? @naive pond level 1 !!! 
how can i get drive links of ai models?
Ayo? @pine mango level 1 !!! 
Yes.
-colab
Suggestions for @crisp plover
- AICoverGen-NoWebUI, by Ardha Google Colab
- RVC Disconnected, Kit Lemonfoot Google Colab
- easyGUI, by rejects Google Colab
- RVC-HF, by r3gf Huggingface Spaces
- AICoverGen, by r3gf Huggingface Spaces
- RVC v2 Huggingface version, by Clebersla Huggingface Spaces
- Advanced RVC Inference, by neuclya Huggingface Spaces
You can find more info on the #1159513888199540817 channel. If you can't find your answer, feel free to ask for help in #✨│ai-help. Credits to Faze Masta and Antasma for compiling these links.
rvc disconnected
you dont have a gpu
and rvc gui its a lil bit old i think
you can use your cpu
i dont suggest you to do inference with that tho
amd/intel inference is also a thing
Traceback (most recent call last):
File "/home/user/app/app.py", line 300, in vc_single
audio_opt = vc.pipeline(
File "/home/user/app/vc_infer_pipeline.py", line 542, in pipeline
pitch, pitchf = self.get_f0(
File "/home/user/app/vc_infer_pipeline.py", line 319, in get_f0
self.model_rmvpe = RMVPE(
File "/home/user/app/rmvpe.py", line 334, in __init__
ckpt = torch.load(model_path, map_location="cpu")
File "/home/user/.local/lib/python3.10/site-packages/torch/serialization.py", line 815, in load
return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
File "/home/user/.local/lib/python3.10/site-packages/torch/serialization.py", line 1033, in _legacy_load
magic_number = pickle_module.load(f, **pickle_load_args)
_pickle.UnpicklingError: invalid load key, '<'.
Traceback (most recent call last):
File "/home/user/.local/lib/python3.10/site-packages/gradio/routes.py", line 523, in run_predict
output = await app.get_blocks().process_api(
File "/home/user/.local/lib/python3.10/site-packages/gradio/blocks.py", line 1440, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/home/user/.local/lib/python3.10/site-packages/gradio/blocks.py", line 1341, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/home/user/.local/lib/python3.10/site-packages/gradio/components/audio.py", line 349, in postprocess
file_path = self.audio_to_temp_file(
File "/home/user/.local/lib/python3.10/site-packages/gradio/components/base.py", line 325, in audio_to_temp_file
temp_dir = Path(self.DEFAULT_TEMP_DIR) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'```
can anyone help me to understand why am i getting this error on Ilaria RVC (rmvpe only)
Can someone help me set up to where I can make my own original ai songs? I have my model just lost
how many epochs should i set for three wav files on rvc disconnected
how many minutes?
are they
Ayo? @hard flame level 1 !!! 
shoot i dont really know if rvc disconnected can do it
ok
it has a max of 4hours runtime
oh'
maybe i should learn how to train locally
you can try but i dunno if it will work
#1159513888199540817 here you can find whatcu need
thanks tho
The conda Installer for RVC studio isnt doing anything but showing me a empty cmd HELP?????????????
i posted this yesreday and no one helped
What interface/fork are we talking about here
The conda Installer for RVC studio isnt doing anything but showing me a empty cmd HELP?????????????
3rd times a charm
Honestly I'm
Not sure
Im using RVC and im trying to use it in a gradio. live
Ayo? @worthy glade level 1 !!! 
Hey can anyone help me? I wanna upload a model to huggingface so that I can use it in Ilaria_RVC but I can't get the right link anywhere.
google colab give me runtime disconnetedes
im using The Mangio-RVC-Fork and for some reason whatever i do when i start training it gives me this error each time ``` File "C:\Users\NOPE\Mangio-RVC-v23.7.0\runtime\lib\site-packages\torch\nn\modules\conv.py", line 309, in _conv_forward
return F.conv1d(input, weight, bias, self.stride,
RuntimeError: cuDNN error: CUDNN_STATUS_NOT_INITIALIZED
also im using a GeForce MX330 gpu if thats the problem
does anyone know what is singer id?
Ayo? @prime helm level 1 !!! 
As far as I know you can just ignore that
i need help
how can install all of this stuff
numba==0.56.4
numpy==1.23.5
scipy==1.9.3
librosa==0.9.2
llvmlite==0.39.0
fairseq==0.12.2
faiss-cpu==1.7.0; sys_platform == "darwin"
faiss-cpu==1.7.2; sys_platform != "darwin"
gradio
Cython
future>=0.18.3
pydub>=0.25.1
soundfile>=0.12.1
ffmpeg-python>=0.2.0
tensorboardX
functorch>=2.0.0
Jinja2>=3.1.2
json5>=0.9.11
Markdown
matplotlib>=3.7.1
matplotlib-inline>=0.1.6
praat-parselmouth>=0.4.3
Pillow>=9.1.1
pyworld>=0.3.2
resampy>=0.4.2
scikit-learn>=1.2.2
starlette>=0.26.1
tensorboard
tensorboard-data-server
tensorboard-plugin-wit
torchgen>=0.0.1
tqdm>=4.65.0
tornado>=6.2
Werkzeug>=2.2.3
uc-micro-py>=1.0.1
sympy>=1.11.1
tabulate>=0.9.0
PyYAML>=6.0
pyasn1>=0.4.8
pyasn1-modules>=0.2.8
fsspec>=2023.3.0
absl-py>=1.4.0
audioread
uvicorn>=0.21.1
colorama>=0.4.6
torchcrepe
Alooo Ben Yavuz İnce İzmir/Buca Sevgi Ferit Akın Ortaokuunun öğretmeni
Ben müzisyenimdir ayrıca sigara içmeyi severim bekarım
Ama beni yaşatan orman cinlerinden çok ama çok teşekkür ediyorumm alooooooooooo beni dinle alooooooo ben çocukları severim ama ben öküz insaları sevmem bunu biiniz
@half cove thoughts??
‼️‼️
hitler
Bru
RVC is a piece of software for making the covers
help i keep getting an error no matter what model and audio file i use on ilaria
What error?
Error None?
Are you logged in huggingface?
Yes!
Ayo? @flat girder level 1 !!! 
I'm sorry I don't understand? Duplicate what space? I don't see a purple button
Or use this one for the time being https://huggingface.co/spaces/r3gm/Aesthetic_RVC_Inference_HF
If it keeps getting an error duplicate it too
ooh ok ill try that
Good luck!
After duplicating on both of them, ilaria keeps giving me an error and aesthetic keeps processing & counting endlessly
Am I doing something wrong? Could it be my computer?:(
2-3 mins
But it does it with the other audios that other people had uploaded in the space too when I tried to choose them
I also think I tried with audio files that were like just a couple seconds to test it
Yeah because you didn't duplicate that yet, correct?
Can you screenshot your settings?
Ayo? @flat girder level 2 !!! 
try converting with dio
that seems to have worked! Thank you!!!!!
np!
@proper shale hey uhm..
The reason I'm not looking for songs. I'm trying to do voice lines.
You know?
Like help me
?
I don't really understand what you wanna do
Do you want a fork for voice line stuff?
So there's this video that has alot of voice lines.
Each sentence, etc.
And I want that to go to the voice model
So you wanna convert it to another person?
Or do you wanna make your own voice model
does anyone know any effect on adobe pr to hide the the ai voice it has like abit noise at the end of every sentense
how many minutes of audio should i have for applio
using uvr to seperate vocals
It's better to have good quality 5 mins sometimes than bad quality 10 mins
ok
Ayo? @hard flame level 2 !!! 
Good luck!
ty
Np :)
FR Easy- 9/17 its not working can someone telll me why
how can i remove people talking in the background, i already used UVR didnt do much?
\
Convert it to another person
what model did you used?
im creating a dataset
yh i mean what UVR model did you used
all of them 
jk just the ones recommended in the doc
whats the name
they worked ig but it didnt remove background talking lol
like u can still hear people speaking in the background with the main vocal intact
so my dataset can be .flac right
waw is better
also which is better apolio, rvc or colab?
oh ok, the document says flac is better but idk if it would work cuz i usally use wav
yh same
wb this
rickie youre clutch, thats all i have to say.
They shouldpromote u to helper
or
i mena
helper king
nvm lmao theres no top helper ig
if u have a good gpu (3060 min) i suggest local
you dont have limits on local
if ya aint a goof gpu you need colab
the disconnected one
ummm apparently i have a shit gpu but a good processor based of userbenchmark
ig im using my baby collab
thanks rickie ❤️
Don't use UserBenchmark😭
They are so biased towards Intel
oop i have an intel i7 lol
hmmmm considering their bias....
also some 1974 gpu....
they prob right
so whats the different between easy 10/25 collab and the offline RVC
bruh whats the point of using offline RVC if tensorboard doesnt work...
did u run feature exctraction??
also yes i did do feature extraction
mh
try to restart runtime and restart "set training variables cell" and "index training cell"
how do i restart runtime?
yh
then this
thanks
does anyone know how to fix this error on the amd version of applio
Rename files/folders so that there's no spaces
now i get this error
when i run rvc i get running with system python nie moPress Any Key To Continue
why i always get error yesterday everything worked
i do with huggingface space
instantly error when converting
ilaria's rvc??
i tried 3 diff rvcs
RVC-HF, by r3gf
Huggingface Spaces
RVC v2 Huggingface Version, by Clebersla
Huggingface Spaces
what problems did you have??
yesterday everything worked download model refresh then put file in refresh convert and waited 30secs now its giving me error when pressing convert usedd different browsers factory resetted everything restarted my mac
i had same problems few hours ago
im having trouble trying to open rvc can someone help me pls
i cant open the actual apllication
why is hugging face keep saying error everytime I try to generate 😭
does anybody know how to fix the error in UVR saying no such file or directory for the reformatted wav (or mp3) file? it should be making it and putting into the temp folder or whatever while in use but i can't seem to figure the issue. have done model inference already, but trying to start using uvr to get some good vocals to train a dataset
Same with me
Anyone have a fix?
where is it better to do covers?
every huggingface says error for me for some reason and others are having same issue is huggingface just not working?
Ayo? @brittle wing level 1 !!! 
gonna keep plugging away... this is the exact error btw FileNotFoundError: [Errno 2] No such file or directory: '/home/anon/codeman/ai/rvc-webUI/TEMP/alright.wav.reformatted.wav'
having same issues
Anyone know why my audio would be sounding this way? Doesn't matter what voice model I use. The output ends up super garbled.
Ayo? @agile kiln level 1 !!! 
Just came back to RVC after a few weeks of not using it and not sure what happened to have broken things like that
?
Only thing I can think of is updating MacOS to Sonoma
that actually was solution i think. weird
whats next best before rmvpe?
that would be mangio crepe right
only thing i can think of is its potentially a permission issue? (Ubuntu 22.04)
Huggingface rvc keeps saying "error" everytime I try to generate!! Is there a solution? 🫠🫠
Oh so the rmvpe is not working
to everyone it looks like rmvpe is not working on hf spaces so if ya got any error try to use mangio-crepe or wait for the problem to be fixed

huh, we've had a few users get that recently
Not sure what's up with it
Nothing significant was changed though... Really weird
beats me, checked permissions and everything thats the think i could think. its not creating the file in the temp folder clearly
everything else works, can't complain too much 🙂
i could never get
illaria to work
i always get errors
i would try converting a song
it never loas
once i refresh everything errors
not even downloadable models
its pretty much impossible for me to make covers
don't use spaces in model names...
Anyone getting errors ?
what u redownloaded? ITS a website ?
i mena the model
Where can i convert my mp3 into ai getting errors on huggingface spaces even after reboot
Ayo? @robust wind level 2 !!! 
Error preprocessing data... Make sure your data was preprocessed.
should i just ignore this lol
cause the extract feature cell on collab shows green check mark
BRUHHHHHH when i click train model its not working
im getting errors
Enter your ngrok authtoken to open tensorboard. Get one here: https://dashboard.ngrok.com/get-started/your-authtoken WHAT DOES THIS MEAN
why i keep getting errors rn? alternative spaces?
ME2
PINNED MESSAGES
idk mate spaces are having a lot of issues tonight
please check out #📰│dev-updates
Does anyone know why when I make a model locally it sounds like static noise/ringing?
Do smaller datasets usually need more epochs?
What is the best way to split back vocals from main ones
I'm using mangio btw
Your dataset is clean, right?
No background noise nor harmonies
Let me try something real quick
Nvm it was because I didn't have the pretrained models installed
Should be okay yeh.
I'd start at 300-500 but you can always stop training at any point if you notice it overtrain in the tensor graph.
do voices with effects the raw voice to train work or not
is there any other alternative
to training likw 20 mins of dataset
and for it not turning off cuz of google colab time restriction
I tried to install the Mangio-RVC-Fork but when I try to run infer-web.py I get this error :
envs\RVC\lib\site-packages\gradio\components.py", line 33, in <module>
from ffmpy import FFmpeg
ImportError: cannot import name 'FFmpeg' from 'ffmpy'
Any idea on how I could fix it ?
ImportError: cannot import name 'CSVutil' from 'my_utils'
anyone know a fix for this?
this is local EasyGUI for RVC v2
Ayo? @gaunt ruin level 2 !!! 
anyone know how to fix this?
try running app.py
Same error :/
Field required [type=missing, input_value={'data': [], 'event_data'...on_hash': 'mtou3jvqd9m'}, input_type=dict]
anyone know how to fix this?
default is fine, only time I've noticed a difference was setting it to 0.99 when it trained for longer with negligible improvement or even slight overtraining
or setting it too low caused it to undertrain
other than that set training epochs to 9999 so it has time to stop itself
Field required [type=missing, input_value={'data': [], 'event_data'...on_hash': 'mtou3jvqd9m'}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.3/v/missing
Pleaseeee anyone help me???
don't know what that is sorry
applio was installing then my pc restarted
Ayo? @pale rock level 1 !!! 
does it suppose to do that?
What are the best setting for a taking and for a singing model to make it realistic and not kinda robotic? I'm using Applio and I have a 4060 ti (16GB) and a 8min dataset.
I don't believe so
just try to install again?
I would say so
ill try it
if you want to be extra safe then delete any files it made already and start from scratch
yeah doing it
it shouldn't go for this long... I also solved the previous issue, but now not even the console shows anything
it opens visual studio installer normal?
ah all good
@somber parrot do you have any advice?
go to http://localhost:7897/
Ayo? @somber parrot level 5 !!! 
thank you bot
I just reinstalled python and all pip packages and nothing changed, any idea on that ?
#✨│ai-help message
you can download the runtime here https://huggingface.co/IAHispano/applio/tree/main
if you have nvidia then it's runtime.zip
extract it to your applio folder
no idea, I use applio
what does applio let you do? does it let you train models and use them?
yes
or is EasyGUI just better in that case?
preference
extract it here or the other applio folder
where can I download Applio?
put runtime in that folder
I found it to be a pain in the butt to install the first time but I'm using it now so ¯_(ツ)_/¯
Ayo? @pale rock level 2 !!! 
last question
could you word that differently?
go on?
can it be distorted or have any effects on it
if that's your desired voice. I made a few robot voices like that such as Bastion or Echo from Overwatch
I'm not doing so then
I already suffered that with easygui
can i link the voice and see if it would work
sure
https://www.youtube.com/watch?v=PmFCgdx9l8U i want to try this
but idk if it would work
understandable lol
only think I would say is remove the reverb, it should still sound robotic like Kassadin but cleaner
can i remove reverb in audacity?
use uvr
Ultimate Vocal Remover works magic
it has a model to remove reverb
is there any other alternative
to training likw 20 mins of dataset
and for it not turning off cuz of google colab time restriction
yah what's the best locally run version
the thing is that he has so few voice lines will even 50 sec work
I just installed applio and it had the same error 😢
Ayo? @remote basin level 1 !!! 
I think you could check the wiki for more voice lines
its all the voice lines, they repeat
are you on linux?
on different occassions
No I'm on w10 and Python 3.10
applio is trash (mostly)
use an alternative fork
you could try it
I tried Mangio earlier but I got an error
dunno then
python is a pain in the butt
then try using standart rvc
standard
This does work bc it doesn't use ffmpeg
but I wanted to try a fork
Ok I found a fix https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/commit/2edeb7168b18f54cabe9902fee010c7926c027f1
Installing ffmpy==0.3.1 rather than the one that is in the requirements.txt seems to work
ERROR: Could not install packages due to an OSError: [WinError 5] Access is denied: 'C:\Users\OneDrive\Desktop\Desktop_Folders\AI_Voice_Models\LOCAL_RVC_V2\Mangio-RVC-v23.7.0\runtime\Lib\site-packages\~aiss\_swigfaiss.cp39-win_amd64.pyd'
Consider using the --user option or check the permissions.
anyone got a fix?
how do i use uvr to remove reverb
go to settings and download reverb hq under MDX-Net in the download center
I think it's because of OneDrive
found a tutorial that says to use vr arch deecho dereverb
is that other option better?
you can try that too but I found reverb hq to work better at least on robot voices
ok
see what works for you
fuck
worked perfectly
Ayo? @pale rock level 3 !!! 
this need to work in linux
... too much shenanigans goin on in windows
just checked, it's not the problem
PLEASE HELP
Try running as admin then
how much time does it usually take for a 200 epoch training?
which one does live voiceover?
fuck
depends on dataset, batch size, and gpu
okada
3050 4 batch size with 50 secs of dataset shouldnt take long
should be fast
the only thing that takes time is the saving proccess
my rvc keeps disconecting me
also if ive done the install and downloaded fthe model once do i have to do it again
does train feature index take long
No
Like less then 10 seconds
It’s training the model that takes long
Run go-web or easygui on admin?
Ayo? @gaunt ruin level 3 !!! 
is there a local filepath for where models are stored on applio rvc? I'd like to mass import all my on system models from a different rvc to applio
logs
thanks alot with the help the final result is just perfect
RVC Guides (How to Make AI Cover)
Translation by country
No
No
You shouldn’t be using kits in the first place
RMVPE Problem on spaces should be fixed
Ayo? @merry roost level 1 !!! 
does batch size really matter for the quality of the models?
Yes
bro 😭
where
Whats tgat
How do you know about this server if you dont know what rvc is
i just joined from yt
nah you're gonna get bullied so hard 😭
What video
some ai voice covee of a song
Ayo? @ebon obsidian level 1 !!! 
i just thought it was cool 😭
Then can you take a guess what rvc does?
nope
wow i am stupid
Or changing someones voice to another persons in general
Yeah check #1159949278270193734 if you wanna use rvc stuff

Batch size refers to the amount of data processed at once (speed option, not quality). This depends on the GPU's VRAM. For example, on an RTX 2070 with 8GB of VRAM, you would use a batch size of 8.
On a Colab GPU, 20 is the value that people consider safest to avoid errors, but I've also heard that it's better to stick to a power of 2 (so 2, 4, 8, 16, 32). Therefore, I use 16 on Colab.
i read the aihub guide website but it sounded like it wasnt really well known... is there any source where i can find how setting a batch size to the power of two works better
I am not sure. I use colab and usually if dataset length < 10m = batch size 8 and if dataset length > 10m = batch size 16
Everytime I try to run rvc it just closes application
i see :o thanks. guess ill have to experiment 😞
guys, how many hours of speech is needed for a good dataset? like 1-2 hours?
wanna train ai
around 20-30 mins of dataset from my experience can be 100, i guess you can try 100 or higher?
no idea
😔
hi i try to convert locally but it give me this error
Traceback (most recent call last):
File "C:\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\routes.py", line 437, in run_predict
output = await app.get_blocks().process_api(
File "C:\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1349, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "C:\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\blocks.py", line 1283, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "C:\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 2586, in postprocess
file_path = self.audio_to_temp_file(
File "C:\RVC\Mangio-RVC-v23.7.0\runtime\lib\site-packages\gradio\components.py", line 360, in audio_to_temp_file
temp_dir = Path(dir) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'
someone know how to fix it?
Ayo? @raw narwhal level 1 !!! 
that's normal, leave it
Okeyyy
fck im blind XD
Np 😂
?
RVC Guides (How to Make AI Cover)
Translation by country
the fields NEVER stop refreshing, any ideas @tranquil cliff or any other helper?
and the command console just stays there
doing nothing
what is that?
local?
yeah, local mangio RVC
Sorry, I'm not very knowledgeable about rvc local
do you know anyone who is?
@low shard
hello
Ayo? @unique palm level 1 !!! 
I using a gradio for trying to make Ai voice.
and I want to include a voice model created by someone else.
but I don't know how to do
I would be grateful if you guys could help me
Nope not me,i do it online as my pc isnt as much powerfuk for local,sorry
You need to download said model on your computer
yes, I downloaded in my computer and unzipped it
but I don't know where to put it...
how do you do it? I've been trying to use my models to produce AI covers, but where do I make them? so far, the only thing that online colab lets you do is train a model, not use it
@molten pecan
so first, click inference or what it is in Korean, download the specified model, place it's .pth in weights and .index in logs folders, and refresh on the gradio, and click your model's pth in your gradio
You can use your models online too, https://discord.com/channels/1159260121998827560/1167844377046040769
@unique palm
really??? finally, jeez, been trying to find a way to do it, thanks.
Your welcome,if theres anything else tell me
thk. like this?
only .index in logs, .pth in weights folder
how do you open the console in Ilaria? you know, to see if the program is making progress?
just wait
RVC doesn't really give updates on inferencinf
yh
oh ok
thanks to your kindness
Does the name of the pth file in the weight file matter? Is it okay to save it as model.pth only?
maybe you should name it as the character of your model
^
is it normal for it to take so long? I mean, I'm fine with waiting, but I wanna make sure I'm waiting for something lol
Yea its normal
alright, I'll trust
Can't upload any files to google colab anyone know why
what colab r we talking about here
and why you should upload files on colab??
What's the best tutorial these days for creating a voice model?
heres on with colab
but if ya got a good gpu like a 3060
you can do it localy
a.. Hello. I made a model sucessfully, and I encountered another problem.
lemme see if i can help you
hmm
Ayo? @unique palm level 2 !!! 
nonetype generally happens when something wasn't selected properly
I only controled pth, index file, and a button named crepe
is there anything I must control
u need to select ur audio
from the dropdown
or insert its path
depending on the GUI you're using
ah
I already put this
and I also confirmed that the route was only in English
is there anything else must to do?
Can you send screenshots of the GUI?
okay
ah
Yes
is my VRAM too small?
@proper shale do ya know what this means?
how much VRAM do you have?
hello! may i ask if anyone has a suggestion of models (male, deep voice, [maybe vtuber/hololive but its fine if not])
6 Gb
Not really, sorry :(
sry to send korean letters. It's my task manager.
maybe not for crepe
Any background task running, @unique palm?
Or a GPU intensive program running, other than RVC?
Try using RVMPE for your pitch extractor
First of all, there were no programs running in the background. But I know how to view CPU usage by app, but how do I view GPU?
What is the difference between an rvc and rvc 2?
RVC v2 is faster and better in quality
ah... I have only three buttons, pm, harvest, and crepe.
for training ? enough
oooh tough question, im not sure
old gui maybe... anyways, try using pm
#1159513888199540817 can help
okay
Thank you ❤️
Ayo? @wraith yew level 1 !!! 
thanks you very much
I'll try that
Can I post a video link here of a RVC tut and someone verify if anything's missing? I followed step by step but it doesnt work for me. Vid is from 5 months ago
and also this long text:
Should be as simple as:
Record your voice for like 15 mins
Get the audio file and trim silences on Audacity
Run through Adobe Enhance if it isn't clean enough
Zip it as a folder with your audio in it, make sure the name is the same as the zip
Go on the RVC V2 Disconnected Colab on #info
Make a new folder in your drive named "rvcDisconnected", upload your dataset zip there.
Install dependencies (first step)
Set training variables - make sure that the character name variable is the same as the folder name in the zip. Don't change CPU threads and Speaker ID. Everything else should be customizable. It's up to you to decide. I recommend using 32k on sample rate, rvmpe as your pitch extractor and keeping the hop rate at 64.
Now run preprocessing and feature extraction, should take a few minutes.
Skip the steps that tell you to resume
Train the index first, scroll down to it a bit, you'll find the step.
After it's done - scroll back up and configure batch size, total training epochs and save frequency. Keep save frequency at 10-20, total training epochs should be 100 in a first session, and batch_size can vary a lot, but 8 should be really good for practically everything
Wait until it's done (Takes about an hour or something, keep looking at Colab page from time to time)
After it's done (has a green tick below the play button), run the export model step, and it should export all your model's stuff - recommend freeing up some space on Google Drive though, it will take 1,5 GB per model
Download the added index and the pth with your model's name on it - zip it, and you have a model! Upload it to Huggingface to upload to our server if you want to - else, try a Google Drive export.
After an upload to Google Drive (if you wanna keep it private), make absolutely sure that the zip is with the right sharing configurations - everyone with the link should see it. After that, just use it with RVC-HF (on #1159949278270193734), go on Resources, Download Model, go back to Inference and be happy and enjoy your model.
Banned
Okay that's way too old, so yeah definitely won't work anymore
dm me for any problems
Unless it's using the Mangio fork locally, then I think it should be alright?
Should I make strange sounds with my voice that RVC usually breaks up on?
But if I train it only on speaking/reading won't it be bad for singing?
...go for it, but don't focus on it
I don't know if I can retrain the model on more data afterwards
i made a Steve Jobs model and he sings alright
Can you give me some tips? Im trying to train using an RTX card and it doesnt seem to be covered very well unless Im missing something
Yes, but it'll have to be an entirely new model
no that ll be fine
Yeah but then I should probably just have more varied audios for the intial training. I'll try making a prototype with just speaking
Since I'm not a local user, I don't believe I can give really good advice, sorry :(
If we're talking about training quality maybe I can help.
Banned
😦
Banning YOU
Thanks I appreciate your answer. I have a HQ audio dataset. About 30 mins long. Just need some guidance on best steps to train
AI hub will be real in 257 minutes....
thats a good idea
I believe you've already trimmed silences and all that, right?
but how i said early speaking/reading should be ok
- making sure it had no reverb and echo
Actually I didnt. Theres silences but I can easily remove them
Ayo? @indigo forum level 2 !!! 
You definitely should. It will improve the model
NoCash will send media in exactly 74 seconds
0,05 secs (for both duration of silence and how short it should be) should work just fine
like nooo spaces at all...when the model speaks he pauses sometimes, is that ok?
Even 0,1 secs
Ayo? @quaint minnow level 2 !!! 
Told ya
never actually gives anything
Those can kinda damage the model
So you need to remove them
It’s a test

