which had a bunch of wrong stuff
#✨│ai-help
karmic locust
copper pebble
nicee!!
karmic locust
but i never explicitly put it there, must've slipped through somewhere along the line
and it made my whole computer think my pytorch was super out of date
and this persisted through uninstallations of python
copper pebble
yh cuz it said that xformers couldnt work with ur python vers or smt like that
copper pebble
xFormers was built for:
PyTorch 2.0.0+cu118 with CUDA 1108 (you have 2.1.1+cpu)
Python 3.9.13 (you have 3.9.13)
lmao
copper pebble
if u need help ping me even tho im not really that good with python
finite galleon
how are these 2 different?
I am assuming you are using aicovergen
brittle wing
yup
finite galleon
Pitch change changes the vocal pitch by an octave
brittle wing
Pitch change changes the **vocal** pitch by an **octave**
i heard someone say male to female should be 12
finite galleon
Pitch change all changes the vocal and instruments by a semitone
finite galleon
i heard someone say male to female should be 12
1 octave = 12 semitones
So on pitch change it needs to be -1 or just 1
brittle wing
hmm
copper pebble
So on pitch change it needs to be -1 or just 1
i dont really suggest pitch changing btw
finite galleon
i dont really suggest pitch changing btw
What
Wdym?
You like male singing high notes?
copper pebble
dunno everytime i tried to change pitch i just ended up with a shitty audio
so i prefered to find the male/female version of the song
maybe someone that did a cover of it
finite galleon
dunno everytime i tried to change pitch i just ended up with a shitty audio
Shitty in like what way
copper pebble
and i didnt even put -12 or 12
copper pebble
U using aicovergen?
no
molten pecan
pitch was set to 0 so idk
Whenever you do inference, most of the time you won't get a great result on the first try
finite galleon
Oh so overall rvc
molten pecan
Especially if you don't tweak any setting
molten pecan
So take your time to change with the settings and see which one sounds the best
finite galleon
how many models do u have emp??
I kinda stopped posting lol
I am making private models
tranquil cliff
hi
molten pecan
Hai
upbeat talon
Thanks but I don't need to make audio based on a ready-made model
I need to MAKE this model
copper pebble
Thanks but I don't need to make audio based on a ready-made model
I need to MAKE...
whats the problem u encountered
upbeat talon
thanks
finite galleon
Have any of yall tried formant shift?
upbeat talon
whats the problem u encountered
I need instructions on how to create a voice model using RVCHFpriv
I do not know the sequence of actions
tranquil cliff
Have any of yall tried formant shift?
What do you mean?
tranquil cliff
ohhh
onyx rover
@brittle wing, please say, what notepad do you use?
finite galleon
notebook*
He is using the aicovergen colab
Pretty sure its in #1159949278270193734
onyx rover
He is using the aicovergen colab
Oh, ok, thanks
brittle wing
my model is shit again i think
lmao
oh well
here it is
(shit)
whys it so high pitched
supple ibex
why does the sound of the model change with a very minor pitch change
non copyright
copper pebble
why does the sound of the model change with a very minor pitch change
cuz with pitch changing ure pratically changing the intonation
copper pebble
thats why i dont reccomend pitch changing
.
supple ibex
I didnt change the pitch
my question is why does the voice change with a slightly deeper voice
molten pecan
my question is why does the voice change with a slightly deeper voice
That's normal
You need to change the pitch until it sounds however you want it
supple ibex
wont that change the octave and make it sound off key
molten pecan
wont that change the octave and make it sound off key
If you want the pitch to be higher then raise it
It will sound natural don't worry
signal drum
hello everyone
signal drum
laptop
signal drum
Ok .. what about specs?
windows 11 pro, 11th gen
latent kettle
windows 11 pro, 11th gen
Which processor? Do you have a powerful GPU ( Graphic card) ?
And what about RAM ?
signal drum
Which processor? Do you have a powerful GPU ( Graphic card) ?
no... its a basic laptop for daily routine use
latent kettle
no... its a basic laptop for daily routine use
Okay! So you have to use cloud computing!
signal drum
And what about RAM ?
8gb
latent kettle
8gb
Can you show me a screenshot of your specs ! By right clicking on this pc> properties
Do fast! ...
gentle summit
what does epochs mean?
glad zealot
its basically how many times it gets trained
most of the time 300 epochs is decent enough
brittle wing
hey when i speak i hear nothing on dc or other games i need help
stiff goblet
Is there any colab to use VR Arc models like deecho- dereverb or denoise?
onyx rover
@glad zealot I am sorry if Im annoying but why does covergen in collab just duplicate voice 2 times sometimes? I recorded my speech and send it there and on result I got sounds which looks like the devil
But usually it works normal with songs. 50/50
glad zealot
no idea that never happened to me yet perosonally
onyx rover
Do you have notebook which basicaly transform audio without uvr and this new technologies?
rare gobletBOT
Ayo? @onyx rover level 3 !!! 
glad zealot
nope
onyx rover
Oh, ok. Anyway thanks
summer perch
I need help here...
upbeat talon
Hi again
I create a model
AICovergen does not accept zip as a model what did I do wrong?
onyx rover
Hi again
I create a model
AICovergen does not accept zip as a model what did I d...
Try to download some model and then see will it work
upbeat talon
Try to download some model and then see will it work
And how do you make a zip archive in a Huggiface link?
upbeat talon
Try to download some model and then see will it work
A other model works
onyx rover
And how do you make a zip archive in a Huggiface link?
Does your zip include .pth and .index files?
upbeat talon
Does your zip include .pth and .index files?
No((((
rare gobletBOT
Ayo? @upbeat talon level 5 !!!
proper shale
And how do you make a zip archive in a Huggiface link?
-huggingface
azure marshBOT
Suggestion for @upbeat talon
upbeat talon
I dont have
The .pth file of the model
and
The .index file of the model
onyx rover
No((((
Then, in weights folder you will find .pth, in logs folder you will see added...index
proper shale
If you trained on RVC Disconnected, it should be on your Google Drive.
upbeat talon
If you trained on RVC Disconnected, it should be on your Google Drive.
What if i dont have pth file?
But have index
upbeat talon
so u lost ur pth
content/Mangio-RVC-Fork/logs/Nade/3_feature768
Proper Numpy version detected.
(28119, 768),721
Training index...
Adding...
Saving files to Drive...
All done! Your index file has completed training.
I have a index
and log files
proper shale
yeah but did you train model
onyx rover
He thought this work is so fast.
rare gobletBOT
Ayo? @onyx rover level 4 !!!
proper shale
Oh it happens haha
onyx rover
I tried covergen but it has some bugs for me. Is here other notebooks? More simple
azure marshBOT
RVC Guides (How to Make AI Cover)
stable flint
hello, if a vocal conversion has audio thats ducking, how can i fix this with the settings?
proper shale
thats probably ur original audio
soft stratus
@proper shale how i pitch voice in RVC to Neo Cortex, Crunch and Dingodile voice
proper shale
Use the transpose option
soft stratus
@proper shale is it in Ilaria RVC
proper shale
Yeah
proper shale
It's basically set to 0
U can change it to negative if you want a deeper voice
Or increase the number if you want a higher pitched voice
acoustic pollen
i got a problem with the res ms when i start the rvc it goes up to like 30k or more
proper shale
What GPU do you have
acoustic pollen
rx 590
rare gobletBOT
Ayo? @acoustic pollen level 1 !!!
acoustic pollen
i changed from rvc to beatrice and it works ;-;
acoustic pollen
i guess
proper shale
what version are you using
acoustic pollen
1.5.3.16 directML i wathced tutorial ind the guy said its for amd
or smth like that
proper shale
yea thats right
acoustic pollen
sshould i download the 17 ver?
proper shale
nah no need
acoustic pollen
good
proper shale
what f0 were you using with rvc tho
acoustic pollen
crepe_tiny
proper shale
hmmmm id recommend using rvmpe_onnx
acoustic pollen
i tried
proper shale
even after exporting ur models to onnx? hmmm
acoustic pollen
u mean from the models i imported to rvc
proper shale
yeah
acoustic pollen
so model is the .onnx and index like it was?
proper shale
yeah
acoustic pollen
how dom iknow it
rare gobletBOT
Ayo? @acoustic pollen level 2 !!!
acoustic pollen
if i choose the wrong one
snow pasture
My next model.
radiant flare
does RVC have a non-gpu pack i can use?
acoustic pollen
i tried everything and nothing works
silver anvil
plush ocean
Just wondering, this channel is for Realtime Voice Changer, yeah?
sand iris
I mean
Yes and no
Generally #🔍│help-w-okada is the place for that
But there's also the real time voice changer from the RVC project which is not as used
Either channels are fine, it's just that it avoids burrying others' support requests
plush ocean
Ah, gotcha. I'm just getting a hell of a lot of latency and wondering if I've done something wrong or if a 6800XT isn't powerful enough to run it. I got the DirectML version (which I think is the version that doesn't rely on CUDA?)
plush ocean
Not really, but lets be honest they don't sound amazing aha
My buffer is around 130ms, but the res goes to 1000+ with the sample models
The issue is that on custom models it goes to 15000+. Am I just using models that are too high quality?
rare gobletBOT
Ayo? @plush ocean level 1 !!!
fringe glen
is there another way to train voice models without the google colab, because it seems that google colab doesn't give enough free runtime to finish training the voice model?
https://colab.research.google.com/drive/1TU-kkQWVf-PLO_hSa2QCMZS1XF5xVHqs?usp=sharing
^ive been trying to use this colab and it doesn't work
proper shale
#1159949278270193734, rvc disconnected
proper shale
The issue is that on custom models it goes to 15000+. Am I just using models tha...
need to export the models to onnx
plush ocean
I will find out how to do that
molten pecan
https://rentry.org/egirlmodel
Works wonderfully. Thanks!
unkempt hamlet
Guys, i just installed applio and downloaded "trapstar" model
rare gobletBOT
Ayo? @unkempt hamlet level 1 !!!
unkempt hamlet
and i downloaded acapella of travis scott goosebumps, just for test
but when i tried to do the steps, the final result was the exactly travis voice
didnt change anything
gilded willow
i just did the voice ai in google collab but how can i talk with it in discord ?
gilded willow
not rn but how ?
proper shale
you gotta use that colab to use it on discord calls n stuff
its in the pins of #🔍│help-w-okada
and u need an ngrok account and a virtual audio cable
gusty kestrelBOT
half cove
gilded willow
and u need an ngrok account and a virtual audio cable
i did all of that but idk how to link it with the discord app because that's in browser
rare gobletBOT
Ayo? @gilded willow level 1 !!!
proper shale
i did all of that but idk how to link it with the discord app because that's in ...
just use the vb Cable as input on discord
and as output in W-Okada
or whatever virtual cable u might be using
gilded willow
just use the vb Cable as input on discord
i'll try tomorrow, tysm
idle pulsar
#1159514067187277865 eren
hot lagoon
What's Recomendation segment size in uvr??
This what I have so What's Recomendation segment size for me>
coral thistle
hello
don't know if im a chimp, but i get this error, and cannot for the life of me figure it out
shoot
weak wren
does anyone know why replay is skipping most of the song? and not doing the whole song?
coral thistle
don't know if im a chimp, but i get this error, and cannot for the life of me fi...
C:\Mangio-RVC-v23.7.0>C:\Mangio-RVC-v23.7.0\runtime\python.exe EasierGUI.py --pycmd C:\Mangio-RVC-v23.7.0\runtime\python.exe --port 7897
Traceback (most recent call last):
File "C:\Mangio-RVC-v23.7.0\EasierGUI.py", line 28, in <module>
from my_utils import load_audio, CSVutil
ImportError: cannot import name 'CSVutil' from 'my_utils' (C:\Mangio-RVC-v23.7.0\my_utils.py)
C:\Mangio-RVC-v23.7.0>pause
Press any key to continue. . .
so, with that i can't run easier gui
what could i do
coral thistle
hello?
rare gobletBOT
Ayo? @coral thistle level 1 !!!
finite galleon
<@818050831034613771>
Pitch changing from female to male yeah i think needs to be reduced
But from male to female nope
Wait actually.
Nah reduce everything
copper pebble
lmao
finite galleon
Lowering lowers the sharpness and highering highers the sharpness which makes it high pitched
6 or -6 should be fine
As long as the instruments follow the key
finite galleon
the problem comes there
Just lower it in pitch
copper pebble
Just lower it in pitch
yh but most of times doesnt sound that good imo
finite galleon
yh but most of times doesnt sound that good imo
I mean
Its better than the vocals being in another key
sand iris
<@310840858767523846> I got it working
Oh great. I was just thinking that I still hadn't done anything lol
sand iris
it needs to run with full precision
Fuck
plucky geyser
dude i have been banging my head trying to figure this out, I was ready to give up man
and this is something I've known since I've used stable diffusion before on rocm
same solution
i just forgor
sand iris
Wait did you need to ?
I ran it on half just fine
Stablediff I mean
The plot thickens
plucky geyser
yeah, it generates a black square or grayscale without it iirc
maybe it's not a problem now
finite galleon
going to bed btw gn <@818050831034613771>
Late 13 minutes but gn
plucky geyser
for anyone searching in discord, to get training to work on amd/rocm you need to run with full precision
sand iris
I'll give it another shot hopefully tomorrow
On something like a 10s dataset of something highly recognizable like I don't know, a sound effect
plucky geyser
for anyone searching in discord, to get training to work on amd/rocm you need to...
edit the 48k etc configs and set fp16_run to false
sand iris
And see if it matches
plucky geyser
im using tensorboard to see if it's training
for the longest time the spectrogram has been completely blank, now it actually shows the voice
and some of the graphs have been a straight line, now it's all doing stuff
the only problem is rocm being slow until randomly running at full speed
static gyro
to test Compute units on google i have purchaced 100 units
so i need a lot of people handing me requests
sand iris
im using tensorboard to see if it's training
Supposedly, AMD pulled an nvidia and released a generation with botched fp16
And it might just be the 5xxx cards
sand iris
Is there one last thing you could do? I don't have a python environment available atm
But the issue could be that we're overriding to a newer target than the card you have
What does torch.cuda.get_arch_list() return ?
If there's any targets from the gfx1000 family that are gfx1010 or older, it could help to change the HSA override
static gyro
what woul make training go faster?
plucky geyser
If there's any targets from the `gfx1000` family that are gfx1010 or older, it c...
I changed it to 1030 already if that is what you are suggesting
I've heard you can get 1010 to work on rocm 5.2 though
sand iris
Yeah 1030 is a supported target
the 5700xt is 1010 though
I'm afraid overriding to a newer HSA might just be breaking
desert pebble
which voice model on here has the most fire? I wish there was a way to sort via the amount of fire.
plucky geyser
it's a hacky way to fix it, other 5700xt users have to do it as well
but it is working from what I can tell
sand iris
Significantly slower unfortunately
plucky geyser
I also need to use an older pytorch version to get this to work, but with 1010 on 5.2 supposedly I can use pytorch 2.0
pytorch 2.0 will give me an error if I use it atm
sand iris
It does ?
plucky geyser
yeah
sand iris
IIRC that's what RVC requests by default
plucky geyser
right
if I'm feeling like it I will try 1010 on 5.2 later
but this has been my headache for the past few days so I'm just happy to finally see something different
plucky geyser
Supposedly, AMD pulled an nvidia and released a generation with botched fp16
and if this is true, what performance benefit could I get from running 1010 on 5.2 anyways?
modern oxide
what woul make training go faster?
making the dataset smaller
sand iris
and if this is true, what performance benefit could I get from running 1010 on 5...
Running half precision ?
modern oxide
buying a more powerful gpu
sand iris
Which would straight up divide vram requirements
And accelerate speeds because there's less data to move around
static gyro
making the dataset smaller
what about raisning the gpu threads?
plucky geyser
I thought it's messed up though?
rare gobletBOT
Ayo? @plucky geyser level 5 !!!
sand iris
I don't know if the override does it
plucky geyser
ah
sand iris
Or if it's the implementation
static gyro
like i want all the power used not just some of it
modern oxide
what about raisning the gpu threads?
you mean cpu threads?
static gyro
fell like im waste ram
plucky geyser
alright well I will try it at some point
static gyro
you mean cpu threads?
yeee
sand iris
I'd hope it's the override, because if it isn't, and at some point they officially supported that card, that would be wack (but not really surprising)
plucky geyser
would be nice to run with fp16
azure marshBOT
RVC Guides (How to Make AI Cover)
oblique bone
Hey, i could probably find this info somewhere else but this community is so active
and i would probably find my info faster by just asking
so uh
ive trained before but i dont remember how to find the model after training
im training locally
any help would be great
radiant flare
having successfully used huggingface.co online tools twice,
i got a really basic question
what is the name of the tool / segment that does the
revoicing an audio file?
fleet tusk
paper granite
IDK bro lol
plucky geyser
@sand iris can't install rocm 5.2, gfx1010 does not work on the latest
looks like I have to use full precision unfortunately
brittle wing
The audio is REALLY stuttery and bad and ngl idk what to do
violet heron
The audio is REALLY stuttery and bad and ngl idk what to do
What f0/pitch extraction are you using
brittle wing
What f0/pitch extraction are you using
idk
idk if i am in the right channel ngl
but I can't post images/videos
violet heron
idk if i am in the right channel ngl
You are unless your using a real time voice changer
brittle wing
You are unless your using a real time voice changer
im using "voice changer client demo"
violet heron
im using "voice changer client demo"
brittle wing
<#1159290161683767298>
ty
rare gobletBOT
Ayo? @brittle wing level 1 !!!
oblique bone
Hey, i could probably find this info somewhere else but this community is so active
and i would probably find my info faster by just asking
so uh
ive trained before but i dont remember how to find the model after training
im training locally
any help would be great
violet heron
Hey, i could probably find this info somewhere else but this community is so act...
Check your weights folder for the pth and the logs for index
oblique bone
thanks
fleet yacht
i cant find the google collab page. how to i use the models?
rare gobletBOT
Ayo? @fleet yacht level 1 !!!
sturdy crater
I have index files but none of them are showing in the dropdown. My index files are in the same folder as where the models are stored in RVC-beta\weights
Why could this be?
violet heron
I have index files but none of them are showing in the dropdown. My index files ...
Index files go into the logs folder
sturdy crater
Index files go into the logs folder
cool, thanks
marble jewel
hi, Just wondering if there was a good setting combination people would recommend when rendering out clear vocals that are a bit more consistent. What I am seeing is sometimes it flip flops from high and low pitches. I know i'm limited to the quality of my vocals and a dataset, but just wondering.
molten pecan
i cant find the google collab page. how to i use the models?
-colab
azure marshBOT
Suggestions for @fleet yacht
tepid cave
inf loading thing whenever i try to switch it rvcv2 it just does this and breaks (train thing)
plush compass
Is there any RVC Colab available for learning?
rare gobletBOT
Ayo? @plush compass level 4 !!!
azure marshBOT
👇 Suggestions for @whole violet!
violet heron
Is there any RVC Colab available for learning?
-colab
azure marshBOT
Suggestions for @plush compass
plush compass
What is the difference between RMVPE and RMVPE_GPU?
quick ledge
What is the difference between RMVPE and RMVPE_GPU?
search on google
frank kraken
Hi, Can anyone help me with creating my own rvc model, im finding difficulties since google collab has banned webui for its free users.
azure marshBOT
RVC Guides (How to Make AI Cover)
snow pasture
I could probably guess why left 4 dead models weren't made yet...
snow pasture
Hi, Can anyone help me with creating my own rvc model, im finding difficulties s...
Best option for training.
https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS?authuser=7#scrollTo=NYDdsNhx4Qdi
(You could use extra accounts to bypass the limit and it'll take month to refresh the used up account to use again)
Best option for making AI Covers and getting Separated vocals, backing, and instrumentals
https://colab.research.google.com/github/ardha27/AICoverGen-NoUI-Colab/blob/main/CoverGen_No_UI.ipynb#scrollTo=B3BqnUoePVGd
(You can keep making and generating Covers once everythings complete by switching the urls of the songs after it finishes)
rare gobletBOT
Ayo? @snow pasture level 10 !!!
copper pebble
cool, thanks
pth goes in weights and index goes in logs (only added index its requested)
sand iris
Does anyone know how to install pytorch
Why do you need that ? It should be handled for you
queen gust
Why do you need that ? It should be handled for you
Wdym?
sand iris
Without using CMD?
Also no, unfortunately it'll require it at some point
queen gust
??
sand iris
If you use any of the installers/releases you shouldn't have to do a manual download
Of pytorch
sand iris
Oh yeah
You don't need to install anything beforehand for voice-changer
This has a section with direct download links to the right versions. All that's needed is a single zip
(and one other software if you want to use the modified voice as microphone)
queen gust
(and one other software if you want to use the modified voice as microphone)
I got it downloaded
But the thing is when I open
Start_https
\Downloads\MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.17\MMVCServerSIO>MMVCServerSIO.exe -p 18888 --https false --content_vec_500 pretrain/checkpoint_best_legacy_500.pt --content_vec_500_onnx pretrain/content_vec_500.onnx --content_vec_500_onnx_on true --hubert_base pretrain/hubert_base.pt --hubert_base_jp pretrain/rinna_hubert_base_jp.pt --hubert_soft pretrain/hubert/hubert-soft-0d54a1f4.pt --nsf_hifigan pretrain/nsf_hifigan/model --crepe_onnx_full pretrain/crepe_onnx_full.onnx --crepe_onnx_tiny pretrain/crepe_onnx_tiny.onnx --rmvpe pretrain/rmvpe.pt --model_dir model_dir --samples samples.json
It comes up with that
On a cmd page
queen gust
https://rentry.co/VoiceChangerGuide
Is that the same for
sand iris
it is
sand iris
It comes up with that
anything else happens? Let it be for a while, it's downloading stuff
Sometimes it doesn't display the download bars for some reason
queen gust
anything else happens? Let it be for a while, it's downloading stuff
Yeah it says like a buch of stuff and then instantly closes
sand iris
Weird
Other possibility is adding one line to the start_http script
Open it with any text editor (notepad works) and add
pause
At the end of it
It will make the window remain even after an error
queen gust
Open it with any text editor (notepad works) and add
```
pause
```
At the end ...
How do I do open with
sand iris
"edit"
queen gust
rare gobletBOT
Ayo? @queen gust level 2 !!!
queen gust
So I just add pause at the end right?
sand iris
yeah
sand iris
Looks good I think
queen gust
Looks good I think
FileNotFoundError: Could not find module 'C:\Users\Bok26\Downloads\MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.17\MMVCServerSIO\sklearn.libs\vcomp140.dll' (or one of its dependencies). Try using the full path with constructor syntax.
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "MMVCServerSIO.py", line 18, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "voice_changer\VoiceChangerManager.py", line 9, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "voice_changer\Local\ServerDevice.py", line 13, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "librosa_init.py", line 211, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "librosa\decompose.py", line 19, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "sklearn_init.py", line 80, in <module>
File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
File "<frozen importlib._bootstrap>", line 1006, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 688, in _load_unlocked
File "PyInstaller\loader\pyimod02_importers.py", line 385, in exec_module
File "sklearn_distributor_init.py", line 21, in <module>
File "PyInstaller\loader\pyimod03_ctypes.py", line 79, in init
pyimod03_ctypes.install.<locals>.PyInstallerImportError: Failed to load dynlib/dll 'C:\Users\Bok26\Downloads\MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.17\MMVCServerSIO\sklearn\.libs\vcomp140.dll'. Most likely this dynlib/dll was not found when the application was frozen.
[19772] Failed to execute script 'MMVCServerSIO' due to unh
queen gust
Looks good I think
Still don't work ;-;
queen gust
You unzipped the full folder right?
Yep
sand iris
Yeah looks good
Is your computer locked down or something? Those errors rarely happen and I'm not sure how to fix them
queen gust
: )
queen gust
Is your computer locked down or something? Those errors rarely happen and I'm no...
Locked down?
sand iris
Like are functionalities disabled by someone else ?
sand iris
Really not sure then, sorry
queen gust
oof
signal drum
Hi guys... i tried using this google colab but again and again i'm facing error... could someone please help me... https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS
rare gobletBOT
Ayo? @signal drum level 2 !!!
signal drum
dreamy flume
My Inferencing voice in Applio is empty despite there being a pth file in the weights folder.
rare gobletBOT
Ayo? @dreamy flume level 1 !!!
dreamy flume
Going to try installing Windows since I'm on Line Ucks (Linux)
sand iris
Going to try installing Windows since I'm on Line Ucks (Linux)
Should work though
dreamy flume
Should work though
Doesn't for some weird reason.
I have my models in their right folders.
sand iris
No errors or anything ?
dreamy flume
No errors.
dreamy flume
I don't think it does detect the models.
Only the index.
Well, I think I'm going to install Windows anyways.
Nvidia drivers on Linux suck so much.
sudden socket
is the download a one time thing or do you need to wait for the app everytime you open the app
sand iris
one time thing
sudden socket
and do you need to acess the cmd prompt everytime you open it
@sand iris i apologize for the ping but can i dm you?
sand iris
A screenshot ? Sure
sudden socket
yeah thanks
sand iris
You should get perms in a bit to send them here
but yeah while you don't, sure, go ahead
sudden socket
thanks
ancient phoenix
If you’re trying to make an ai singing cover, is it better to train the voice with singing or talking
rare gobletBOT
Ayo? @ancient phoenix level 1 !!!
proper shale
Singing, because you have more diversity in terms of pitch
Talking models can sing, but in a very limited range.
proper shale
State your problem
pearl basin
Hello. Can you help, nothing is loading
quartz frost
I need help to convert a song, its more than 4:10 which is impossible for my google colab to convert
is it ok?
i can send the song and the voice model
please?
azure marshBOT
copper pebble
@quartz frost use the 1st one here
quartz frost
...how come is it impossible to convert?
my pc cant use rvc so i got a google golab for it
but it disconnects for songs more than 4:10
copper pebble
my pc cant use rvc so i got a google golab for it
but it disconnects for songs m...
just use the 1st link on this embed or Ilaria rvc
i think u were using something pretty outdated
quartz frost
i think u were using something pretty outdated
proud elbow
If you’re trying to make an ai singing cover, is it better to train the voice wi...
singing parts can have more pitch range and talking parts can have more phoneme range, so you can include both as long the voices sound same
copper pebble
this is the one i use https://colab.research.google.com/github/ardha27/AICoverGe...
thats should work
if the problem persists just use Ilaria rvc
quartz frost
if the problem persists just use Ilaria rvc
does it also disconnect me if I take too long?
copper pebble
gonna go back in class rn so i wont answer for a while
sudden socket
is there any helper available?
proper shale
State your issue
sudden socket
i started installing the rvc app and its downloading a bazillion things
rare gobletBOT
Ayo? @sudden socket level 1 !!!
proper shale
i started installing the rvc app and its downloading a bazillion things
That is normal.
It's installing its dependencies and models.
sudden socket
this is normal right
proper shale
Yep.
sudden socket
It's installing its dependencies and models.
oh alright i got scared for a bit
cause the last time i did it only installed like a couple of them
proper shale
Oh it's probably not updating the download progress correctly
So it's probably the same 6/7 things iirc
But eh, if it ain't broke, don't fix it
proper shale
Np!
dreamy flume
Well, I think I'm going to install Windows anyways.
finally got it to work, it detects everything, i think it's a wonky linux install
rare gobletBOT
Ayo? @dreamy flume level 2 !!!
sudden socket
i cant hear anything on the app for some reason
proper shale
What are your input/output settings? Send a screenshot
sudden socket
yup now i can atleast listen something ill try to toggle with those settings
jovial tiger
how to i get the ai to work with discord
proper shale
you need a virtual audio cable
wintry latch
so I've been looking online for a solution to this problem but I can't seem to find one, maybe I missed something but I'm having this issue where when I go to first launch the start_http it goes to download the stuff but 3 things keep not downloading, like they either don't get any bytes/sec or they only briefly get maybe a 100-300 bytes, I assume this means it doesn't fully install because when I try any voice my CMD is a loop of "pipeline is not initialized. waiting generate pipeline..." any ideas? ( I did find 1 post that was basically the same issue to mine in regards to the looping but there isn't exactly a clear solution)
jovial tiger
i cant hear any thing or no one can her me when i use the virtual audio cable
rare gobletBOT
Ayo? @jovial tiger level 1 !!!
azure marshBOT
RVC Guides (How to Make AI Cover)
wind arrow
Yeah I'm having a bit of a problem with actually opening RVC. Two days ago it was working fine with connecting to the local host but today it seems like my browser (opera GX) won't let me connect and keeps saying "localhost refused to connect"
carmine ether
Can anyone tell me what RVC is exactly and what it stands for
proper shale
Yeah I'm having a bit of a problem with actually opening RVC. Two days ago it wa...
Are you using start_http
or start_https
proper shale
so I've been looking online for a solution to this problem but I can't seem to f...
Yep, it doesn't install correctly
I'd recommend trying a VPN to Europe/America to get better download speeds, but this will vary with your internet speed
wind arrow
Are you using start_http
I usually just open it by typing localhost:7897 on the website search bar
proper shale
i cant hear any thing or no one can her me when i use the virtual audio cable
Input/output is configured wrong
jovial tiger
how would i fix it?
wind arrow
Opens its own little window
Where do i type it? Command prompt? (I'm pretty new so no idea what I'm doing)
rare gobletBOT
Ayo? @wind arrow level 1 !!!
proper shale
Where do i type it? Command prompt? (I'm pretty new so no idea what I'm doing)
Just open the .bat file in the MMVCServerIO folder
start_http.bat
carmine ether
When I try to select an Input, I always get this error message.
rare gobletBOT
Ayo? @carmine ether level 2 !!!
proper shale
how would i fix it?
In W-Okada, these settings must be like this:
Input: Your microphone
Output: CABLE Input/Line 1 (if using VAC by Muzychenko
And on Discord:
Input: CABLE Output/Line 1
Output: Headphones
proper shale
np
proper shale
When I try to select an Input, I always get this error message.
Hmmm never saw this before
wind arrow
Just open the .bat file in the MMVCServerIO folder
on the RVC beta folder right?
carmine ether
Hmmm never saw this before
My screen before the error message
proper shale
on the RVC beta folder right?
oh wait you're not using W-Okada?
i think it's go-realtime.bat in this case
carmine ether
My screen before the error message
n when I go to input n select smth, it crashes/gives me this error
wind arrow
oh wait you're not using W-Okada?
Nope I'm using RVC
proper shale
Ohhhh I see I see
wintry latch
I'd recommend trying a VPN to Europe/America to get better download speeds, but ...
so is there any options without a vpn? cause the download speed for the rest of the stuff is fine, but it's just those 3 specific items that seems to have a problem with their download?
carmine ether
My screen before the error message
wait isnt this okada already?
proper shale
so is there any options without a vpn? cause the download speed for the rest of ...
Don't think there is another solution, sorry
carmine ether
What GPU do you have?
Me?
proper shale
Yes
carmine ether
proper shale
you have the wrong version for your GPU
wintry latch
Don't think there is another solution, sorry
ok, so I'm trying to install it rn on the main drive at the root home and for some mythical reason they are downloading just fine? hopefully this means it's fixed? fingers crossed
rare gobletBOT
Ayo? @wintry latch level 1 !!!
proper shale
ok, so I'm trying to install it rn on the main drive at the root home and for so...
yooo i think it is
hmmmm thats weird
carmine ether
you have the wrong version for your GPU
what do I do then
carmine ether
uninstall everything Ive got so far?
proper shale
yeah cuz it's the version for amd
carmine ether
ohh I see thanks
proper shale
np
wintry latch
yooo i think it is
ok, it's passed the first lot of downloads, it has gotten to "downloading model files" and it downloaded 2 with ease, but there is a 3rd which has paused at the beginning and seems to be tweaking out, hopefully if it doesn't install this one correctly it should be fine since it's simply a model file, right?
proper shale
I hope so 🙏
wind arrow
i think it's go-realtime.bat in this case
i finally solved it. Thanks!
proper shale
Yay! Happy to have helped ;)
wintry latch
I hope so 🙏
'Connection Broken' I'm in tears
wintry latch
I'm just gonna attempt to open the start_http again and hope and plead that it isn't something I needed
carmine ether
yeah cuz it's the version for amd
Isnt the one u sent amd aswell?
proper shale
no, its the nvidia version
carmine ether
oh okay im just confused bc it says amd64
and yea after opening it I get the same error message as before
wintry latch
I'm crying and shaking rn
proper shale
😭
carmine ether
and yea after opening it I get the same error message as before
after clicking on Input
proper shale
and yea after opening it I get the same error message as before
ohh god I don't know actually
carmine ether
Is it maybe smth with my settings? Idk
rare gobletBOT
Ayo? @carmine ether level 3 !!!
carmine ether
I can change output without any effort tho
But as soon as I change my Input to anything (I tried all) I get the error
wintry latch
Now I have a theory..... a game theory......
proper shale
I can change output without any effort tho
But as soon as I change my Input to ...
@sand iris got any idea of what's up?
sand iris
Nothing in logs ?
wintry latch
I've deleted the "models" folder and am restarting it, and it seems to be re downloading the thing it didn't last time, so if this finishes, that should be everything (meaning if it fails after this I am gonna give up)
sand iris
That happened on your custom models or also the sample ones ?
sand iris
<@310840858767523846> got any idea of what's up?
maybe connecting via web browser would help ?
carmine ether
If anyone knows a fix, pls help. Happens every time I try to select an Input
brittle wing
How to reduce ms??
carmine ether
smaller chunk
wintry latch
OMG I GOT IT TO WORK I AM POGGING!!!
proper shale
YESSS
proper shale
How to reduce ms??
use a faster f0 like rvmpe
brittle wing
My res up almost 30000 ms
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
help me
brittle wing
Hmmm what GPU do you have
My GPU is a GTX1650
proper shale
Are you using the version of W-Okada with the GPU0-3 buttons?
brittle wing
no
proper shale
Okay, so you're in the right version
Try using crepe_tiny as your f0
And not use index and keep extra low
rich scaffold
I load Retrieval-based-Voice-Conversion-WebUI to run in local pc and I got this error
2023-11-28 20:31:43 | ERROR | asyncio | Task exception was never retrieved
future: <Task finished name='2v5x2f4usjp_16' coro=<Queue.process_events() done, defined at E:\RVC Memory Project\venv\lib\site-packages\gradio\queueing.py:343> exception=1 validation error for PredictBody
event_id
Field required [type=missing, input_value={'data': [0, './audios/so...on_hash': '2v5x2f4usjp'}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing>
Traceback (most recent call last):
File "E:\RVC Memory Project\venv\lib\site-packages\gradio\queueing.py", line 347, in process_events
client_awake = await self.gather_event_data(event)
File "E:\RVC Memory Project\venv\lib\site-packages\gradio\queueing.py", line 220, in gather_event_data
data, client_awake = await self.get_message(event, timeout=receive_timeout)
File "E:\RVC Memory Project\venv\lib\site-packages\gradio\queueing.py", line 456, in get_message
return PredictBody(**data), True
File "E:\RVC Memory Project\venv\lib\site-packages\pydantic\main.py", line 164, in __init__
__pydantic_self__.__pydantic_validator__.validate_python(data, self_instance=__pydantic_self__)
pydantic_core._pydantic_core.ValidationError: 1 validation error for PredictBody
event_id
Field required [type=missing, input_value={'data': [0, './audios/so...on_hash': '2v5x2f4usjp'}, input_type=dict]
For further information visit https://errors.pydantic.dev/2.5/v/missing
```.Increase chunks to like 128
brittle wing
Increase chunks to like 128
How to turn off playback??
proper shale
Set your output to your virtual audio cable
brittle wing
can someone send me the download link for windows 10 with a GTX 1650 super?
brittle wing
can someone send me the download link for windows 10 with a GTX 1650 super?
???
download link to what
the rvc program
wheat mirage
you mean drivers?
brittle wing
you mean drivers?
i dont really now haha i have a .pth and .index file and want to play a little bit with real time voice cloning
earnest dune
carmine ether
If anyone knows a fix, pls help. Happens every time I try to select an Input
Anyone knows where I could find some help to my issue?
soft stratus
@proper shale this RVC i use it and how i find transpose options from these choose
proper shale
<@622856015444049937> this RVC i use it and how i find transpose options from th...
scroll up
soft stratus
@proper shale do you mean this
rare gobletBOT
Ayo? @soft stratus level 6 !!!
proper shale
<@622856015444049937> do you mean this
yes, pitch
soft stratus
@proper shale what i choose right pitch to get Crunch, Neo Cortex Portuguese voice
proper shale
it depends on the voice model and the audio u wanna convert
sharp glen
window
sonic basin
sage creek
how do i find my index?
proper shale
probably in your rvcDisconnected model folder
sage creek
(i'm local)
plush compass
https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw How do I load the dataset from this colab? I tried creating a dataset folder, but it failed... TT https://docs.aihub.wtf/v/korea/undefined-1/model-training-rvc It's not written here either.....
sage creek
let me find it rq
proper shale
https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw How do...
This is an outdated Colab
proper shale
go on Logs
sage creek
ty
plush compass
This is an outdated Colab
I got Easy Colab from Discord. Is there a newer Colab other than this?
proper shale
Yes - RVC Disconnected in #1159949278270193734.
plush compass
Ah thank you so much TT
proper shale
;) np
carmine ether
Okay so I made some progress. None of the Input methods work except for file and system(only win)
What does the last one mean?
proper shale
i think thats using your system's default input
carmine ether
which one would that be on a computer??
brittle wing
is there a way to increase the volume I found it quite quiet
rare gobletBOT
Ayo? @brittle wing level 2 !!!
carmine ether
is there a way to increase the volume I found it quite quiet
Increase gain out
carmine ether
which one would that be on a computer??
apparently its nothing? Atleast I didnt get it to give me any sign of life
So yea appart from file I dont think I can do anything for some reason
proper shale
huh, that's weird
carmine ether
if anyone has Ideas on how to fix it, or needs additional information to help, pls let me know
plush compass
RVC Colab batch size is 16? Or 20? Which do you prefer more?
proper shale
RVC Colab batch size is 16? Or 20? Which do you prefer more?
8 is generally the best
plush compass
8 is generally the best
Is 16 better if there is a lot of data? The data is about 40 minutes.
I've heard that batch size doesn't have much of an effect other than speed... Other people say it has a direct effect on performance, but I'm worried because different people say different things.
proper shale
Would be a bit faster but would also be worse per epoch than 8 batch_size
proper shale
I've heard that batch size doesn't have much of an effect other than speed... Ot...
Yes, it does affect quality.
plush compass
Thank you so much for the advice. TT I started with batch size 16, stopped it after 9 seconds, and trained it again at 8.
plush compass
thank you
proper shale
np ;)
if you have any more questions, ask here, me and other helpers will try our best to answer asap
plush compass
Thank you so much TT
proper shale
:) glad to help
carmine hearth
My question may seem very dumb, but is there any inference colab that has an option to not separate instrumentals?
copper pebble
My question may seem very dumb, but is there any inference colab that has an opt...
-colab
azure marshBOT
Suggestions for @carmine hearth
copper pebble
My question may seem very dumb, but is there any inference colab that has an opt...
the 1st one here
proper shale
nah use the one in pins
rare gobletBOT
Ayo? @proper shale level 52 !!!
proper shale
the no webui version aint workin anymore
brittle wing
i need help
when i try to do the vocal instrumental separation
well
i would link a pic but i cant
carmine hearth
How do I make sure that colab doesn't separate the instrumentals?
proper shale
it just doesn't, by default
brittle wing
??
brittle wing
OneDrive/שולחן: No such file or directory
proper shale
so its a complete ai cover
proper shale
OneDrive/שולחן: No such file or directory
take that folder out of onedrive
brittle wing
i deleted onedrive from my computer
rare gobletBOT
Ayo? @brittle wing level 1 !!!
brittle wing
yet that still happens?
lmfao
it happens because you deleted onedrive
weird bug
tf
i fixed that issue by not having onedrive in the first place at all 😎
private version of windows?
its not private
it costs a lot to get though
carmine hearth
Sorry, I'm using machine translation, so I guess my intentions were misrepresented. I already have a separated vocal file, and I'm afraid it will get corrupted in the separation process inside the inference colab. Is this not a big problem in most cases?
brittle wing
my dads work gave him an enterprise visual studio account
and thats where i have it from
its really good
what stops from pirating
oh wait
C:\Users\baller\AppData\Local\Temp\gradio\d5234f436a4078ca3e66868c38a3366ad2d4d9ce\13: No such file or directory
@proper shale
onedrive doesn't have anything to do with this
but i rlly fucking hate onedrive
carmine ether
Okay so I made some progress. None of the Input methods work except for file and...
I FIXED IT
It had to do with my System Sound Input Settings. I just played around with some settings and now it lets me do everything I want
Why does it cut the audio sometimes?
hot lagoon
in mvsep what the best setings for covers?
rare gobletBOT
Ayo? @hot lagoon level 10 !!!
carmine ether
whoever said that with lag and then deleted
Im having the same issue
brittle wing
How do I reduce latency when recording videos or talking to friends??
merry schooner
I think I have the wrong version of it installed.. could anyone send me a guide or perhaps help me get the right one..?
It's not letting me pick my Nvidia. It just shows gpu0-gpu3.
And then to add on top of that, my res is skyrocketing whenever I say something.
copper pebble
It's not letting me pick my Nvidia. It just shows gpu0-gpu3.
whatcu using??
merry schooner
whatcu using??
using as in like graphics card or
rare gobletBOT
Ayo? @merry schooner level 1 !!!
copper pebble
using as in like graphics card or
i mean the programm haha
carmine ether
It's not letting me pick my Nvidia. It just shows gpu0-gpu3.
use .15 at the end
copper pebble
voice changer??
merry schooner
mhm rvc
carmine ether
Anyone knows how I make it sound less lagy and less like a robot?
merry schooner
use .15 at the end
Is there a download to that?
carmine ether
Is there a download to that?
yes hang on
I have nvidia and thats the link to the one im using
rare gobletBOT
Ayo? @carmine ether level 4 !!!
merry schooner
got it, I'll test that one out and let u kno if it works for me.
carmine ether
sure thing
merry schooner
sure thing
I gotta extract the file right? or am I wrong about that
proper shale
yh
carmine ether
I gotta extract the file right? or am I wrong about that
Yea
carmine ether
Anyone knows how I make it sound less lagy and less like a robot?
Somebody?
violet heron
Anyone knows how I make it sound less lagy and less like a robot?
What f0 pitch extraction are you using?
violet heron
Change crepe tiny to RVMPE
plush compass
What is the save extra files to drive option in the Index Training options?
violet heron
What is the save extra files to drive option in the Index Training options?
Save the extra files that are not needed
carmine ether
Change crepe tiny to RVMPE
It worked! Thank you
About chunk
Do I want it to be low or high?
gusty kestrelBOT
I have found 11 results for the search Adam...
Requested by redeyez6666_37032
violet heron
Its been stuck at this for a while
Wait for it
merry schooner
Wait for it
how long does it usually take first download? It's been like 3 minutes
I dont know if its maybe that I already had the old version installed
but it is taking a little
merry schooner
oh u did ?
carmine ether
yea. I also had the wrong version installed first
I then deleted it and started fresh
indigo herald
/help any tts bots / server
harsh scarabBOT
No command called "any" found.
plush compass
resample_sr , rms_mix_rate . filter_radius What exactly do these three options do?
carmine ether
why does this happen out of nowhere
carmine ether
why does this happen out of nowhere
It was because Chunk was too smal
But what does the number below that do?
"Extra!
"
naive fossil
how to fix?
merry schooner
anyone know what this does?
quartz cairn
Is there working colabs out there?
latent kettle
How can I use UVR 5 on colab ?
hot lagoon
what is KaraFan?
fringe crow
violet heron
/help any tts bots / server
Use /speak
proper shale
anyone know what this does?
volume sensitivity slider
if it reaches below said number it doesn't process that as ai
naive fossil
how to fix?
@proper shale can u help me?
proper shale
Is this really an error
carmine ether
Would the quality be better if I didnt do realtime?
proper shale
Is this really an error
Like is the program working
violet heron
Is there working colabs out there?
Check pins
proper shale
Would the quality be better if I didnt do realtime?
Yes
carmine ether
And the overall quality isnt perfect but Eula is extremely trash for me
violet heron
And the overall quality isnt perfect but Eula is extremely trash for me
So it’s probably the model
proper shale
Probably the model
carmine ether
It has by far the best samples tho so I kinda feel like it has to be smth on my end
Doesnt get much better if I dont do real time either
well okay im gonna give up on that one.
But how do I improve the overall quality?
violet heron
well okay im gonna give up on that one.
But how do I improve the overall qualit...
If your using realtime mess around with the in/out, tune, chunk and extra settings
carmine ether
what does extra do?
violet heron
If your not doing real time make sure everything is as clear as possible
carmine ether
I got the in/out, tune and chunk at the optimal settings already
violet heron
what does extra do?
I forgor
hexed dew
is the rvc disconnected dead for feature extraction?
Im tryna train my index but I cant, will I have to do it locally?
proper shale
is the rvc disconnected dead for feature extraction?
This is a problem with your dataset
hexed dew
dataset is around 1:30 mins long but i've shortened to around 15 9 second clips
i used the dataset maker for another model and it worked fine
proper shale
What "dataset maker"?
hexed dew
Instant dataset maker, it just chops the clips and removes empty space
shadow wigeon
What to do if vol is 0.000 Please help?
proper shale
Instant dataset maker, it just chops the clips and removes empty space
You can use just 1 audio only and it'll work well
No need to chop up in multiple audios
hexed dew
What length do you recommend?
proper shale
RVC does that automatically anyway
shadow wigeon
What to do if vol is 0.000 Please help?
proper shale
What to do if vol is 0.000 Please help?
What does cmd say?
latent kettle
RVC does that automatically anyway
If I put an audio into RVC with pauses.. will it automatically cut that pauses ?
latent kettle
Im tryna train my index but I cant, will I have to do it locally?
What error are you getting?
proper shale
If I put an audio into RVC with pauses.. will it automatically cut that pauses ?
...no
you have to cut those manually
(truncate silence effect in Audacity does that too ;)
shadow wigeon
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pipeline...
rare gobletBOT
Ayo? @shadow wigeon level 1 !!!
proper shale
[Voice Changer] Pipeline is not initialized.
[Voice Changer] Waiting generate pi...
yep smth went wrong with ur initial installation
hexed dew
but i will retrain on a longer audio
proper shale
yep smth went wrong with ur initial installation
Delete pretrain and model_dir and restart start_http.
shadow wigeon
Delete pretrain and model_dir and restart start_http.
go to voice chat
Only I'm Russian
proper shale
can't, sorry
shadow wigeon
okey
latent kettle
its asking if i ran feature extraction
Sometimes it gets mad .. at first.. it also gave me this same error
proper shale
Anyways, @hexed dew, I recommend exporting your audios only via Audacity. Seems like any other dataset exporting method gives errors...
hexed dew
Anyways, <@317661218170994690>, I recommend exporting your audios only via Audac...
good point, I was using FL studio to export
I had better luck with audacity
Let me try it
hexed dew
Sometimes it gets mad .. at first.. it also gave me this same error
how did you fix it?
shadow wigeon
How can I make vol robotic?
latent kettle
good point, I was using FL studio to export
I have also prepared my dataset with fl studio but It worked
hexed dew
90 seconds of audio should be enough shouldnt it?
proper shale
Zipping also seems to be a problem - if you zip it as a .rar, it can screw up.
latent kettle
how did you fix it?
I was just training for experiment purposes just on 10 epochs! So I do it again,
proper shale
Renaming it doesn't make it any better either.
hexed dew
does the runtime even read rars?
proper shale
That's the thing. It doesn't
proper shale
90 seconds of audio should be enough shouldnt it?
I'd recommend getting more if you can
latent kettle
Zipping also seems to be a problem - if you zip it as a .rar, it can screw up.
Yes this factor can be considerable.
proper shale
How much more?
like 10 mins I'd say
latent kettle
and is 200-250 epochs enough for say 2-3 mins of audio?
I don't think.. it's a good option
proper shale
and is 200-250 epochs enough for say 2-3 mins of audio?
Epochs vary a lot, this can only be told by TensorBoard
latent kettle
and is 200-250 epochs enough for say 2-3 mins of audio?
I trained my model on 30 minutes of dataset in 350 epochs
shadow wigeon
How can I make vol robotic?
hexed dew
I trained my model on 30 minutes of dataset in 350 epochs
Alright i will try to get 5 mins of audio and do 100 epochs
rare gobletBOT
Ayo? @hexed dew level 4 !!!
latent kettle
How can I make vol robotic?
Just train a model in a robot voice,
shadow wigeon
Just train a model in a robot voice,
Can you help?
proper shale
How can I make vol robotic?
what do you mean
latent kettle
Alright i will try to get 5 mins of audio and do 100 epochs
If you want to train your model well ... You need minimum 10 minutes long audio clip
hexed dew
It doesn't have to be amazing quality
latent kettle
Can you help?
Do you want to create a voice.. which sounds like a robot?
shadow wigeon
I can't produce a voice where res buf and above vol
proper shale
...well, I'd recommend it being decent quality
shadow wigeon
Do you want to create a voice.. which sounds like a robot?
no
proper shale
I can't produce a voice where res buf and above vol
I'm not sure I understand
hexed dew
...well, I'd recommend it being decent quality
Yes of course but there are limited clips and the ones that are there is all I have
TTS isn't so helpful as it americanises the voice
shadow wigeon
Let me PM you
proper shale
Yes of course but there are limited clips and the ones that are there is all I h...
Try cleaning them up as well as possible, if you can, for the best possible model
hexed dew
I have noise cancelled and de verbed in FL studio
the quality of clips is very good
thats why I used just 1:30 mins of it
latent kettle
thats why I used just 1:30 mins of it
Add more data into it .. make a 10-20 minutes long audio clip
proper shale
oh thats good then
azure marshBOT
RVC Guides (How to Make AI Cover)
proper shale
Add more data into it .. make a 10-20 minutes long audio clip
They can't - they just said that
latent kettle
They can't - they just said that
Ohh ! I didn't read it ...
latent kettle
thats why I used just 1:30 mins of it
So try with more epochs
exotic pond
Are a lot of the available models bad quality? I'm having trouble making any sound good with Mangio-RVC, with the exception of Trump or Obama.
proper shale
Well yeah
exotic pond
Damn, ok.
copper pebble
u can use 11labs or ilaria tts
theres a guide for a 2nd one here #1159513888199540817
11labs give a certain num of words per month
ilaria is infinite and free
copper pebble
11labs give a certain num of words per month
then u have to pay or change account
yes is for mobile but you cant make sasuke speak directly
can first download the tts audio that u get and then convert it on ilaria rvc using sasuke model
pretty easy
proper shale
Not directly, you have to get a TTS result from somewhere else then convert with RVC
copper pebble
like i said before no
just download the audio u get from the tts and then convert it using the model that u want in ilaria rvc
azure tangle
`/content/Mangio-RVC-Fork/logs/Gargarensis/3_feature768
Exception Traceback (most recent call last)
<ipython-input-18-9c42c5055659> in <cell line: 30>()
29 listdir_res = list(os.listdir(feature_dir))
30 if len(listdir_res) == 0:
---> 31 raise Exception("No features exist for this model yet. Did you run Feature Extraction?")
32
33 try:
Exception: No features exist for this model yet. Did you run Feature Extraction?`
Can someone help with this error?
I got it on RVC v2 Disconnected in the Index Training part
proper shale
How was your dataset exported?
rare gobletBOT
Ayo? @remote verge level 2 !!!
remote verge
But not quite yet
ty
proper shale
np
azure marshBOT
This interaction has expired, use the command
/guides realtimeif you wish to see it again.
radiant flare
trying a new (to me) approach, since non of my local attempts have panned out.
using RVC-beta0717 for cpu based local install.
$ ll inf*
infer_batch_rvc.py
infer_uvr5.py
infer-web.py
the guide i have says use infer-web.py (via a bat file)
i would rather use the command line, is one of these other infers gonna be what i want? or is this a waste of time?
rare gobletBOT
Ayo? @radiant flare level 4 !!!
soft stratus
@proper shale what's number of voice pitch to get Saga De Gemeos's (Gilberto Baroli) voice
steady pond
proper shale
<@622856015444049937> what's number of voice pitch to get Saga De Gemeos's (Gilb...
I don't know sadly
But you can find it manually
Just test different pitch values
violet heron
How to resume training
Local or Colab?
noble dawn
Collab
violet heron
There’s some resume training cells
rare gobletBOT
Ayo? @noble dawn level 1 !!!
noble dawn
violet heron
Do you remember your models step count?
noble dawn
500
violet heron
Not epochs, steps
noble dawn
oh nah i dont think
violet heron
Put that in
noble dawn
how do i check
violet heron
I forgot how to 💀
noble dawn
dm me rq
violet heron
Wait till a helper comes they should be able to help
noble dawn
ok
azure marshBOT
RVC Guides (How to Make AI Cover)
radiant flare
does the bot have a guide for doing a local linux install?
sand iris
Unfortunately no
But install should be :
- Get python 3.10, git, c++ build tools (make, cmake, g++, python headers), aria2
- setup a venv
- clone RVC
pip install -r requirements.txt- run the dlmodels.sh script while being at the root of the cloned repo
- run infer-web.py
noble dawn
if i use a differnt collab can i continue training
latent kettle
Tried to run training index again
You can still use your model without index
hexed dew
You can still use your model without index
how?
brittle wing
how do I fix this in applio when i CLEARLY have tensorboard installed??
latent kettle
Just export your model to Google drive by running the cell . Open the RVC . Select your model and use it .
hexed dew
Just export your model to Google drive by running the cell . Open the RVC . Sele...
Im using RVC gui, so the v1 architecture
i run it locally so how can I run it without the index
latent kettle
i run it locally so how can I run it without the index
Just leave the intex empty and convert your audio
latent kettle
Im using RVC gui, so the v1 architecture
Are you trying to train your index locally and getting errors