#✨│ai-help
blazing solar
Im but some part just left me confused
blazing solar
And then click run all?
edgy tangle
wait
first click this
click the "(1):"
an then click the play button that shows up
blazing solar
Ok afterthat continue with the rest?
edgy tangle
yeah, then click the "(2):"
two times to click the play button too
and then click the last one
blazing solar
After that wait for the applio link?
blazing solar
Okay
edgy tangle
when it looks something like this, its ready
blazing solar
Yes
edgy tangle
no problem, try to ask later or wait to someone else that knows about it
blazing solar
Ok
latent cypress
something unexpected always happens when something nearly succeeds
twilit forge
36 epochs
twilit forge
wait it to get at least 50 epochs
yeah
twilit forge
you can do inference while training?
got my friend to do it
edgy tangle
you can do inference while training?
you can stopping training and then resuming it
and yes, you can while training, but it is too slow
latent cypress
okay okay
twilit forge
can i use mangio and applio at same time?
twilit forge
Now what is this👽
click on the link
blazing solar
click on the link
Error
The applio link error
I wish applio google colab is working 😭 kaggle is hard af for me
latent cypress
I wish applio google colab is working 😭 kaggle is hard af for me
yeah it's hard, but at least it faster than colab😭
blazing solar
yeah it's hard, but at least it faster than colab😭
😭Still i prefer gooogle colab
twilit forge
can i use mangio and applio at same time?
?
blazing solar
Now im stuck with this error link
twilit forge
Now im stuck with this error link
just locally train it
latent cypress
😭Still i prefer gooogle colab
i know, cause it's easier to use
blazing solar
Im doomed
blazing solar
just locally train it
How
blazing solar
Is llaria rvc zero also not working?
latent cypress
Is llaria rvc zero also not working?
yes only for inference, not training
blazing solar
Ughhh
simple ore
colab is broken again, the previous uv workaround does not work
brittle wing
Maybe cause I made a copy in my drive
brittle wing
colab is broken again, the previous uv workaround does not work
Again right
simple ore
new fix for Applio colab
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install numpy==1.23.5 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install gradio==5.23.1
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install -r requirements.txt -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install numpy==1.23.5 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install gradio==5.23.1
brittle wing
Can you post it in a web browser...
simple ore
basically add this to all uv install lines
brittle wing
In a separate cell?
simple ore
no, in the install cell
brittle wing
no, in the install cell
Why can't you edit the Colabs
simple ore
brittle wing
Separate cell
simple ore
no
brittle wing
Row by row
simple ore
the whole cell ```# @title Install Applio
!sudo apt update
!sudo apt install python3.10
!sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
!sudo update-alternatives --set python3 /usr/bin/python3.10
!curl -sS https://bootstrap.pypa.io/get-pip.py | python3
import sys
sys.path.append('/usr/local/lib/python3.10/dist-packages')
import codecs
from IPython.display import clear_output
rot_47 = lambda encoded_text: "".join(
[
(
chr(
(ord(c) - (ord("a") if c.islower() else ord("A")) - 47) % 26
+ (ord("a") if c.islower() else ord("A"))
)
if c.isalpha()
else c
)
for c in encoded_text
]
)
org_name = rot_47("Vkkgdj")
new_name = rot_47("kmjbmvh_hg")
uioawhd = rot_47(codecs.decode("pbbxa://oqbpcj.kwu/QIPqaxivw/Ixxtqw.oqb", "rot_13"))
uyadwa = codecs.decode("ncc.cl", "rot_13")
A = "/content/" + rot_47("Kikpm.ovm.bu")
!pip install uv pyngrok
!git clone --depth 1 $uioawhd $new_name --branch 3.2.8-bugfix --single-branch
%cd $new_name/
clear_output()
print("Installing requirements...")
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install -r requirements.txt -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install numpy==1.23.5 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv pip install gradio==5.23.1
clear_output()
print("Finished installing requirements! ")```
for UI colab
similar for noUI
brittle wing
My reaction
brittle wing
the whole cell ```# @title **Install Applio**
!sudo apt update
!sudo apt instal...
Not a separate one?
latent cypress
noobies said the whole cell
brittle wing
noobies said the whole cell
The install cell w additional rows
simple ore
you can either edit the existing cell or make a new one
it is not additional rows
it is just adding parameters for UV
brittle wing
you can either edit the existing cell or make a new one
Edit most likely
brittle wing
it is just adding parameters for UV
?
twilit forge
graph at 91 epochs
brittle wing
the whole cell ```# @title **Install Applio**
!sudo apt update
!sudo apt instal...
If I copy this will it be formatted at different rows.
latent cypress
graph at 91 epochs
is that from your friend?
twilit forge
is that from your friend?
no this is mine
but using the same vocals
im the only one training
latent cypress
no this is mine
lyery said avg loss 50 is more accurate than loss g total
latent cypress
set the smoothing to 0.6
twilit forge
oh yeah thats easier to read
latent cypress
ikr
twilit forge
its at 1.9k now
latent cypress
yeah, just keep training
twilit forge
hopefully this best one cus we have made so many models of it
a lot of em been good
but not all of em are raw vocals
latent cypress
just like you, i've made a bunch of models too, but I never really knew how to train them properly, only recently started learning more about it
twilit forge
this one was from yesterday
just there was bit of noise that fucked it cus of vocal remove
low shard
Is llaria rvc zero also not working?
it works, just that ilaria rvc zero is meant only for inference
latent cypress
just there was bit of noise that fucked it cus of vocal remove
which model did you use to remove the vocals?
brittle wing
the whole cell ```# @title **Install Applio**
!sudo apt update
!sudo apt instal...
How is it for no UI...
twilit forge
which model did you use to remove the vocals?
mvsep
bs melformer
i think sum like that
simple ore
How is it for no UI...
similar change
latent cypress
mvsep
i recommend you to use melroformer voc_fv4 by gabox, this model can extract fuller vocals than the bs model from mvsep, it's old tho
twilit forge
which model did you use to remove the vocals?
i also mixed this with raw vocals
brittle wing
similar change
In what sense also it's hard to edit w a phone keyboard.
latent cypress
i also mixed this with raw vocals
voc_fv4 models can gives you almost like raw quality vocals (imo)
twilit forge
voc_fv4 models can gives you almost like raw quality vocals (imo)
ill try it now rq
latent cypress
ill try it now rq
https://huggingface.co/spaces/TheStinger/UVR5_UI you can find the model here
twilit forge
https://huggingface.co/spaces/TheStinger/UVR5_UI you can find the model here
oh yeah i normally use melband
my bad i forgot
latent cypress
my bad i forgot
np, make sure you used the right melband model, because some of them is separating fuller vocals and generating annoying noise
brittle wing
@simple ore I need applio no UI for training
brittle wing
https://colab.research.google.com/drive/1y42NG3StPnbx_BzhCj94CE4aQCJx8gMR @simple ore done, how do I fix applio no UI?
brittle wing
good night ai hub server
Nite
twilit forge
good night ai hub server
goodnight
brittle wing
```!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit u...
Same replacement withouthe gradito row?
twilit forge
u think if i open mangio while training on applio it will blow my pc up 😭
low shard
u think if i open mangio while training on applio it will blow my pc up 😭
Mangio rvc should be banned smh
I hope you were just kidding since mangio is abandoned since 2023
brittle wing
Mangio rvc should be banned smh
Puoi mettere i link ai colab che ho appena aggiustato a qualche posto?
Solo chiedendo
urban wave
WHERE DO I FIND MY PTH FILE I GOT THE INDEX NOT THE PTH (applio)
simple ore
if you save all weights you'll have the small model saved every time d/g saves
urban wave
its just 1 index file and folders
urban wave
if you save all weights you'll have the small model saved every time d/g saves
IDK HOW TO DO THAT
im on 3.2.8
simple ore
there was a bug that if you chose saving every 100 epoch and max 100 it did not save the model
urban wave
i dont have that option
simple ore
you need to save every x epoch
urban wave
bettt
simple ore
save only latest saves one copy of D/G
save every weight also saves the small model
you may run out of space as D/G are really big
urban wave
betttttttttttttt
rugged crane
for Image models, is there currently a workaround for creating copyrighted characters?
My brothers paid gpt won't create me comics based on celebrities and I need them urgently
that's ridiculous, my career kinda depends on this
simple ore
local install does everything
steel kite
The collab for RVCAICoverMakerUI isn't working
It's just giving this
Traceback (most recent call last):
File "/content/main_program/main_program/main.py", line 1, in <module>
import gradio as gr
ModuleNotFoundError: No module named 'gradio'
twilit forge
@simple ore
unique rock
which ?
simple ore
instead of backend container stuff add UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit uv to uv pip install lines
unique rock
This is inside the applio installation but it still throws an error
unique rock
I'm going to try
humble river
rvc support online?
unique rock
@title Install Applio
!sudo apt update
!sudo apt install python3.10
!sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
!sudo update-alternatives --set python3 /usr/bin/python3.10
!curl -sS https://bootstrap.pypa.io/get-pip.py | python3
import sys
sys.path.append('/usr/local/lib/python3.10/dist-packages')
import codecs
from IPython.display import clear_output
rot_47 = lambda encoded_text: "".join(
[
(
chr(
(ord(c) - (ord("a") if c.islower() else ord("A")) - 47) % 26
+ (ord("a") if c.islower() else ord("A"))
)
if c.isalpha()
else c
)
for c in encoded_text
]
)
org_name = rot_47("Vkkgdj")
new_name = rot_47("kmjbmvh_hg")
uioawhd = rot_47(codecs.decode("pbbxa://oqbpcj.kwu/QIPqaxivw/Ixxtqw.oqb", "rot_13"))
uyadwa = codecs.decode("ncc.cl", "rot_13")
A = "/content/" + rot_47("Kikpm.ovm.bu")
!pip install uv pyngrok
!git clone --depth 1 $uioawhd $new_name
%cd $new_name/
clear_output()
print("Installing requirements...")
UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit !uv pip install -r requirements.txt -q
UV_PRERELEASE=if-necessary-or-explicit !uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q
UV_PRERELEASE=if-necessary-or-explicit !uv pip install numpy==1.23.5 -q
UV_PRERELEASE=if-necessary-or-explicit !uv pip install gradio==5.23.1
clear_output()
print("Finished installing requirements! ")
simple ore
nope
simple ore
and it has to be the same "UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicit" on all lines
unique rock
It worked, thank you very much
humble river
rvc support online?
@simple ore NotADirectoryError: [WinError 267] The directory name is invalid: 'C:\Users\vedant\Videos\sample.mp3'
end preprocess, for first step of training, the directory is valid
simple ore
it ask for a folder name, not for a file name
humble river
it ask for a folder name, not for a file name
im so stupid ty
unique rock
😭 @simple ore
simple ore
show the changes
unique rock
@title Install Applio
!sudo apt update
!sudo apt install python3.10
!sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
!sudo update-alternatives --set python3 /usr/bin/python3.10
!curl -sS https://bootstrap.pypa.io/get-pip.py | python3
import sys
sys.path.append('/usr/local/lib/python3.10/dist-packages')
import codecs
from IPython.display import clear_output
rot_47 = lambda encoded_text: "".join(
[
(
chr(
(ord(c) - (ord("a") if c.islower() else ord("A")) - 47) % 26
+ (ord("a") if c.islower() else ord("A"))
)
if c.isalpha()
else c
)
for c in encoded_text
]
)
org_name = rot_47("Vkkgdj")
new_name = rot_47("kmjbmvh_hg")
uioawhd = rot_47(codecs.decode("pbbxa://oqbpcj.kwu/QIPqaxivw/Ixxtqw.oqb", "rot_13"))
uyadwa = codecs.decode("ncc.cl", "rot_13")
A = "/content/" + rot_47("Kikpm.ovm.bu")
!pip install uv pyngrok
!git clone --depth 1 $uioawhd $new_name
%cd $new_name/
clear_output()
print("Installing requirements...")
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicituv uv pip install -r requirements.txt -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicituv uv pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --upgrade --index-url https://download.pytorch.org/whl/cu121 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicituv uv pip install numpy==1.23.5 -q
!UV_CONSTRAINT= UV_BUILD_CONSTRAINT= UV_PRERELEASE=if-necessary-or-explicituv uv pip install gradio==5.23.1
clear_output()
print("Finished installing requirements! ")
Finished installing requirements!
unique rock
show the changes
still the error appears in start applio
unique rock
Finally, you are my angel, thank you for your patience
simple ore
copy-pasting in colab cells is ass
unique rock
copy-pasting in colab cells is ass
What settings do you recommend for this, after having played the audio?
Or truncated*
runic thorn
Eddycrack864/RVC-AI-Cover-Maker-UI/blob/main/assets/RVCAICoverMakerUI.ipynb
it no longer works, error :
simple ore
Eddycrack864/RVC-AI-Cover-Maker-UI/blob/main/assets/RVCAICoverMakerUI.ipynb
it n...
probably the same uv instal issue
brave swallow
is there any channels got removed in the server lately ?
vestal helm
An error occurred extracting the index: need at least one array to concatenate
If you are running this code in a virtual environment, make sure you have enough GPU available to generate the Index file.
im having this issue with applio
never had it before but now its just suddenly popping up
simple ore
An error occurred extracting the index: need at least one array to concatenate
I...
you dataset is wack
vestal helm
lol
simple ore
too short / slices too big / forgot to slice
simple ore
yes
vestal helm
so what do i do
latent cypress
Finally, you are my angel, thank you for your patience
really happy to see a conversation where people aren’t always trying to be the responders, but actually being good listeners
latent cypress
really happy to see a conversation where people aren’t always trying to be the r...
when someone asks the same thing over and over, it’s not curiosity, it’s just ego talking
brave swallow
I've got keyboard click in this part what I have to use to get rid of it ?
simple ore
the overlaps
latent cypress
where tf did the 4 extra mins come from
from the sliced method
simple ore
no, you need some so it captures the context
vestal helm
ah okay
sour sun
Did the uh, did the making-models channel get deleted?
brave swallow
Did the uh, did the making-models channel get deleted?
it seems like that
steel forge
I've got keyboard click in this part what I have to use to get rid of it ?
Remove that whole piece from your set lul
If you can't get it with declick and stuff without distorting
sour sun
Soo where do we ask questions about model crafting Here I guess?
latent cypress
Soo where do we ask questions about model crafting <:misc_cry:117667469862975097...
nvm
maybe you can ask here
crude flame
Soo where do we ask questions about model crafting <:misc_cry:117667469862975097...
Here or https://discord.com/channels/1159260121998827560/1192011222023950368 here ig
Imo this change was bad
sour sun
All my incredibly specific questions about models and the helpful answers to them
analog obsidian
tbh I like this new channel
knotty moth
@brittle wing refrain from posting promotional or sketchy link in this support channel
brittle wing
Apologies. It was a link to an article in LessWrong about a behavior in LLMs that I am curious if anyone has any information about.
low shard
tbh I like this new channel
🔥
blazing solar
-kaggle
verbal oasis
is Applio working for you guys on google colabs today?
I'm getting gradio error
brittle wing
How to use any ai models?
dry sorrel
Dawg ain't no way I'm linking my phone to nothing. This is whack wtf
knotty moth
Dawg ain't no way I'm linking my phone to nothing. This is whack wtf
you're not forced to use it
dry sorrel
you're not forced to use it
Well what do you suggest big pimpin ?
simple edge
hi , i have a decent voice recording, but its from a phone .. is it possible to run it trough an ai that will make it sound full and like it was on a good mic enhanced with full spectrum frequencies not missing
like its an studio recording
low shard
Guys what is horizen_id
Horizon is a tunnel service
low shard
Why is kaggle error too what am i supposed to use
Elaborate the issue
!give-media-perms 1h @balmy blaze
Elaborate more what you want to do, what's your PC GPU, what are you following, what's the issue
low shard
is Applio working for you guys on google colabs today?
I'm getting gradio error
You need to use the workaround, check #📰│dev-updates
oak shore
What rvc is this using? Can anyone tell me?
low shard
How to use any ai models?
For realtime voice changing? Or for pre recorded audios? What's your PC GPU?
low shard
What rvc is this using? Can anyone tell me?
That seems to be Applio
brittle wing
For realtime voice changing? Or for pre recorded audios? What's your PC GPU?
Realtime
Mobile
oak shore
That seems to be Applio
thank u
low shard
Mobile
You can't on your phone
Even if you use a cloud service (remote good PC), phones lack of a virtual audio cable needed for re routing the modified voice to other apps
@brittle wing do you have any laptop? Thats the super bare minimum for realtime voice changing, even if it's not powerful you can use cloud, as it laptops have a virtual audio cable
Else it's impossible
brittle wing
<@456226577798135808> do you have any laptop? Thats the super bare minimum for r...
Yes
fervent zephyr
Sorry if I accidentally interfered into someone's conversation, but I can't understand one thing. Do anyone know how to make some voice model(for example English one) speak other languages without problems?(without, like, a little bit distorted words and etc.)
low shard
Sorry if I accidentally interfered into someone's conversation, but I can't unde...
Your only options are either train it to that language, or lower the index (lower means it uses less trained accent and uses more accent from the input audio)
fervent zephyr
Your only options are either train it to that language, or lower the index (lowe...
Ok, thanks
brittle wing
I am in a meeting right now. I talk everything after 20-30 minutes.
low shard
I am in a meeting right now. I talk everything after 20-30 minutes.
Alright, @ me when you're free
brittle wing
Alright, @ me when you're free
Yh
low shard
Ok, thanks
Yw
latent cypress
Tell your PC specs
i feel like i was pointed with a gun every time you ask this😭
low shard
i feel like i was pointed with a gun every time you ask this😭
Why 😭
Y'all don't ever check if your PC meets the minimum requirements when you're going to install a program?
It's very needed for AI and Games for example
lofty lichen
why not add separate model making channel under this? useful information was shared on the previous one
latent cypress
Y'all don't ever check if your PC meets the minimum requirements when you're goi...
yes, i checked the requirements before installing rvc
low shard
why not add separate model making channel under this? useful information was sha...
Users were confused on which channel to use and also kept asking in multiple channels, so we merged them all in one
All the info should be already in our Docs https://docs.aihub.gg/
low shard
maybe i feel like I'm being interrogated or something😭
We ask it only because it's needed for us to know if you're PC is powerful enough, it happened many times that users tried to train models on a 10 year old laptop
latent cypress
We ask it only because it's needed for us to know if you're PC is powerful enoug...
yeah i understand
i think i can't joking about this, because my english skill will make it so rude or misunderstanding
low shard
i think i can't joking about this, because my english skill will make it so rude...
Nahh don't worry
I know you're joking lol
knotty moth
hi , i have a decent voice recording, but its from a phone .. is it possible to ...
try AudioSR but still not recommended for making rvc datasets
low shard
Resemble enhanced could be good too for speech
honest sundial
Whats the best channel to creat voice model
low shard
Whats the best channel to creat voice model
Wdym channel?
You need to use a program
What's your PC GPU
clever birch
about w-okada forked and normal w-okada, if i want max extras do i put the forked one to like 5s
and idk what extras does tbh
knotty moth
about w-okada forked and normal w-okada, if i want max extras do i put the forke...
2.7 s unless for weaker gpus
knotty moth
esp for running games
low shard
about w-okada forked and normal w-okada, if i want max extras do i put the forke...
Extra controls quality a bit
Chunk controls delay
low shard
then isnt 5 a good number
It causes more delay for more resources being used, and also cutoff issues
U can also send a screenshot of your wokada settings
low shard
You're settings seem fine, you can also enable force fp32 for more stable models in advanced settings but can cause more delay
Also you should set the chunk to the perf number + 60
You get the perf number when running the program
knotty moth
You're settings seem fine, you can also enable force fp32 for more stable models...
it would be fine for 4080
clever birch
Also you should set the chunk to the perf number + 60
Oh I didn't know that
Cuz like I followed the fork guide
It say xx80 192 chunk
2.7s
I didn't know what chunk was
low shard
Cuz like I followed the fork guide
That's the general suggested one, the chunk depends on which programs u are gonna pair the Wokada with
clever birch
Also what's force fp32
low shard
Also what's force fp32
It uses the fp32 inference mode
For models
Meaning it takes more resources
But it's more stable and slightly better quality
low shard
Thanks for the help
You're welcome
thorn widget
I'm getting ./run-applio.sh: line 3: .venv/bin/activate: No such file or directory when I try to run applio on linux
simple ore
I'm getting `./run-applio.sh: line 3: .venv/bin/activate: No such file or direct...
did you do run-install.sh?
thorn widget
The folder does not contain such a file
simple ore
did it create .venv (the virtual environment) ?
rough reef
how to add another voice to the program?
thorn widget
did it create .venv (the virtual environment) ?
Yes, it did
But it rather already was there
When I unzipped it
simple ore
well, personally I prefer just making an environment using pyenv
and then installing manually
simple ore
yeah, not sure what's the deal with the compiled version
add rocm/bin path to Path env variable
simple ore
Could you please tell me how xd
install python 3.11.11 using that
then just download the source/clone the repository and pip install -r requirements.txt
thorn widget
install python 3.11.11 using that
Do I need exactly 3.11.11?
simple ore
preferably, does not work with 3.12+
thorn widget
Ok, got it
simple ore
if the system uses <3.12 you dont need pyenv
just a manual python -v venv venv will do
thorn widget
I'm installing python 3.11.11 already
thorn widget
then just download the source/clone the repository and ``pip install -r requirem...
Wait, do I need to clone the repo from hugging face or do it in that compiled folder?
simple ore
Wait, do I need to clone the repo from hugging face or do it in that compiled fo...
git clone https://github.com/IAHispano/Applio.git
thorn widget
k
thorn widget
git clone https://github.com/IAHispano/Applio.git
Oh, sorry, I thought, HF was also a repos hub
simple ore
no, we use hf to store big files
thorn widget
Got it
serene horizon
Diffence between an "added" .index and "trained" .index?
knotty moth
"added" one is usable, and another is not
serene horizon
"added" one is usable, and another is not
Makes sense as "trained" is only 2.3 MB and "added" is 91.4 MB.
brave swallow
If you can't get it with declick and stuff without distorting
yep good Idea
serene horizon
"Using HiFi-GAN vocoder
Converting audio 'assets/audios/audio.wav'"
Why does Applio hang here? It never finishes the conversion? @simple ore
The timer just keeps going with nothing happening. (no CPU usuage).
If i try without the index file, it works. With the index file, it doesn't work.
knotty moth
I've got keyboard click in this part what I have to use to get rid of it ?
remove the whole part or try separating using another effective model like mel roformers or bandit v2
thorn widget
then just download the source/clone the repository and ``pip install -r requirem...
done
Do I need to run install and other files then?
I'm getting this error when I'm running run-applio.sh:
Oh, and also this warning:
A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.2.4 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.
Re: mixing models, is it advisable to mix models from different pretrains together, or is that taboo
Like if i have a KLM and an OGpretrain model, does mixing them together degrade something? Or should I only mix two versions of KLM / two verions of OGpretrain
analog obsidian
just try and see if it works good for u
og has less vocal range so idk how it would react being merged with a klm model
hallow thistle
how to add another voice to the program?
Which program? RVC or W-Okada?
!howtoask
blazing solar
-colab
simple ore
Do I need to run install and other files then?
run just 'python' for me please
requirement are supposed to install numpy 1.25.3
it fails with numpy >2.0
brave swallow
remove the whole part or try separating using another effective model like mel r...
got it
if I got dataset 1H length, can I reach something below 27 in training?
simple ore
if I got dataset 1H length, can I reach something below 27 in training?
total g? it is affected by batch size
larger set requires larger batch, thus the value goes up
flint geyser
What seems to be the problem?
flint geyser
which option is the right one
use GUIDE
tribal pivot
do u know waht the download is thp
tribal pivot
use [GUIDE](https://rentry.co/ForkVoiceChangerGuide)
it doesnt show where to download
tribal pivot
Bro i asks if i want to download nvidia control panel
nvm thx
how tf do i use this
brave swallow
total g? it is affected by batch size
my batch size was 12
I'll check total g when I go home
forest ginkgo
how can i use it on collab
im having a hard time getting it to work
please help me
odd shale
ask <@628603171735994379> for Wokada help
@pastel oak is
flint geyser
oh ok
forest ginkgo
hi
low shard
how can i use it on collab
Use what on colab? What do you want to do? What's your PC GPU?
forest ginkgo
is wokada still working on google collab
forest ginkgo
Use what on colab? What do you want to do? What's your PC GPU?
i got a i5 1235U and integrated graphics
w okada definitly wont work nativly on my machine
low shard
i got a i5 1235U and integrated graphics
Damn, I mean it does work on CPU but would be mostly shitty performance
low shard
is wokada still working on google collab
Cloud is having A LOT of issues, cloud is unstable ASF compared to local
Hina mod original wokada is broken
forest ginkgo
Damn, I mean it does work on CPU but would be mostly shitty performance
yeayh i guessed
low shard
yeayh i guessed
Wokada deiteris fork colab works but you get detected on the free tier for using a web UI, which isn't allowed, so it will actually work only if you pay for colab
forest ginkgo
isnt there anyway around
low shard
isnt there anyway around
Kaggle (another cloud service site) Wokada deiteris fork is the only way, which has a harder interface so it's not easy like colab, needs a phone number verification (it's owned by Google), but gives you 30 hours weekly of better GPU compared to colab
There's no other way on cloud for free than this
Else you can pay for colab or buy a good pc
forest ginkgo
i tried verifying my phone number with collab but for some reason i wont recive the verification code
low shard
i tried verifying my phone number with collab but for some reason i wont recive ...
Not colab, I'm talking about kaggle
Those are 2 different cloud sites
forest ginkgo
i ment kaggle sorry
low shard
i ment kaggle sorry
Try contacting kaggle support then
There's no other way for free
No other work around
We even tried encrypting the Wokada deiteris fork colab code, it still gets detected no matter what since months
forest ginkgo
if i try to run it nativly on my muchine will it be atleast usable?
low shard
if i try to run it nativly on my muchine will it be atleast usable?
For discord VC?
forest ginkgo
yeah
low shard
yeah
You could try, it might be decent
Be aware that it could possibly harm your CPU if done for a long time
Tbh just contact Kaggle support, it's free and they will reply to you
forest ginkgo
how long will they take to respond
stiff goblet
hello. what is that mean ? log interval should be equal to step count until that specific epoch? For example if epoch 150 has 2000 steps, log interval should be edited to 2000 ?
low shard
how long will they take to respond
Who knows, maybe a day or some
low shard
thanks for the help!
Let me know for any other issues or questions
low shard
hello. what is that mean ? log interval should be equal to step count until tha...
You need to train till you reach the number of epochs of the save frequency, get how many steps it took that, delete the saved model, and change the log interval to the steps amount
stiff goblet
in this situation, log interval should be 80 ? epoch save frequency is 10
low shard
in this situation, log interval should be 80 ? epoch save frequency is 10
You should have stopped training long time ago, you had to stop it after going over 10 like at epoch 11 when the epoch 10 was saved
Now you are at epoch 310, so you kinda wasted time
stiff goblet
Now you are at epoch 310, so you kinda wasted time
nah that's ok, each epoch tooks only 8 seconds
low shard
nah that's ok, each epoch tooks only 8 seconds
Uhh what is your batch size and dataset size
stiff goblet
Uhh what is your batch size and dataset size
8 batch size and 6 minutes dataset.
stiff goblet
You need to train till you reach the number of epochs of the save frequency, get...
Since every 10 epoch takes 80 steps, log interval should be 80 ? Am I right?
simple ore
a single epoch should be 100+ steps if everything is right.. for a decently good dataset
<10 steps/epoch is atrocity
graph synching only applies to old applio and mainline rvc
no longer needed as it was dumbest idea to begin with
applio now logs every 50 steps for averages and at the end of every epoch
serene horizon
Can anyone tell me how to do inference on a Mac locally?
tropic garden
Can anyone tell me how to do inference on a Mac locally?
Replay. That is what I use on my Mac.
serene horizon
Replay. That is what I use on my Mac.
Thanks. I looked at that, but I couldn't find where to add the .index file.
tropic garden
I zip my models whe I add them to Replay.
tropic garden
Thanks. I looked at that, but I couldn't find where to add the .index file.
You add them under Model +
serene horizon
You add them under Model +
Yes, I found that but it only asks for the .pth.
tropic garden
Do you have the latest version?
serene horizon
Do you have the latest version?
Yes. The old one used to acknowldge that a index was included, but not this one.
tropic garden
It works without the index.
serene horizon
It works without the index.
Yeah, but it doesn't sound right because you need the index to do an accurate conversion.
tropic garden
Index is for the accent.
cinder wyvern
keep getting this when trying to use the google colab
thorn widget
run just 'python' for me please
$ python
Python 3.11.11 (main, Apr 11 2025, 16:11:07) [GCC 14.2.1 20250207] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>
today's colab fix is to add --prerelease if-necessary-or-explicit to the install cell's uv calls
cinder wyvern
today's colab fix is to add ``--prerelease if-necessary-or-explicit`` to the ins...
tried that and now its just been stuck on installing requirements for like 5 minutes
oh wait ignore me it worked
big thank
deep nova
why does start http on my vc client just inastantly close and nothing opens up
kindred hare
how can i get less delay in rvc
viral flint
hey anyone know any tools similar to VASA-1 but not text-to-speech?
something where you could use your own face to animate an image of another face with ai
somber canyon
is there any way i can increase the quality/ realism of the The W-okada Voice changer, my current settings are 5s extra and 130ms. (i dont mind if it takes a little longer, i just want a better quality)
analog obsidian
getting a better model
coarse pier
real quick, is it just me or does realtime just is not good enough for using it passively on lets say, a game match?
i dont think it has to do with my pc specs, it might have to do with my mic quality i dunno
but like, voice will randomly go super robotic out of nowhere
hallow thistle
Both Discord as an app program and web browser would basically work the same. The difference of audio depends on your microphone, a version of W-Okada you're using, and the settings of it.
hallow thistle
why does start http on my vc client just inastantly close and nothing opens up
Are you trying to use the original version of W-Okada? This one is too outdated. There's a better one available.
hallow thistle
how can i get less delay in rvc
There's no audio delay in RVC program, unless you mean by a realtime voice changer, which is W-Okada.
hallow thistle
is there any way i can increase the quality/ realism of the The W-okada Voice ch...
Are you using fork W-Okada on a cloud service like Colab and Kaggle? Because typical extra number setting isn't supposed to be too high if running locally.
kindred hare
There's no audio delay in RVC program, unless you mean by a realtime voice chang...
yes w okada
hallow thistle
yes w okada
Which W-Okada version are you using? And what is your PC GPU?
kindred hare
Which W-Okada version are you using? And what is your PC GPU?
rtx 4060 laptop gpu and latest cuda version
hallow thistle
but like, voice will randomly go super robotic out of nowhere
Audio sounding robotic can happen when you didn't set chunk and extra settings properly according to your GPU.
hallow thistle
rtx 4060 laptop gpu and latest cuda version
The version number of W-Okada you're using, something like 1.x.x.x or bXXX. A "latest CUDA version" doesn't mean it.
hallow thistle
2.0.76-beta
Download and use this better W-Okada instead. https://rentry.co/ForkVoiceChangerGuide#download-nvidia-on-windows
hallow thistle
You're welcome. Make sure you read everything on the guide I sent to you, especially NVIDIA parts. 
kindred hare
oke
kindred hare
You're welcome. Make sure you read everything on the guide I sent to you, especi...
when i launch it from window sit opens up a browser page instead of app
and in that page i cant choose input and output device
they dont show up
kindred hare
and in that page i cant choose input and output device
nvm that
hallow thistle
and in that page i cant choose input and output device
Make sure you have give audio permission to the tab on your browser.
slow ice
wheres the fork ai voice changer?
hallow thistle
wheres the fork ai voice changer?
violet ferry
does anyone here know how you could make okada work with fivem?
pastel oak
does anyone here know how you could make okada work with fivem?
If its a game or anything, then get a virtual cable, all explained on the guide
https://rentry.co/ForkVoiceChangerGuide
cunning matrix
-colab
violet ferry
If its a game or anything, then get a virtual cable, all explained on the guide
...
i have that, but when i get into fivem it's messed up, every other game works fine tho im pretty sure
pastel oak
i have that, but when i get into fivem it's messed up, every other game works fi...
Define "messed up"
violet ferry
like um, you've looked at my forum on ai support right? i listed what happens in there
inner haven
hey i been using the changer but when i listen back i notice it picks up like crackling noise and i tried it with another mic, anyways to fix that?
tribal pivot
how to know if the w okada is safe
lean solar
Is this a good spectrogram? I'm not audio engineer guy
https://i.imgur.com/qRT8MvS.png
pastel oak
my w okada opens the web browser not app
ye thats normal. the program is safe since its open source and its been checked out by experts
we wouldnt recommend a program to 500k users in the server if we didnt think its safe
all the developers are here
tribal pivot
ye thats normal. the program is safe since its open source and its been checked ...
is there anyway to ue it in the ap
low shard
is there anyway to ue it in the ap
don't use it in the app, just use it in your browser
tribal pivot
but wont it lag more
low shard
the great majority of AI programs use a web user interface, original wokada did too, just that it opened it's own window
low shard
but wont it lag more
no, or do you have around an extremely shit ton of tabs opened in your browser?
tribal pivot
no, or do you have around an extremely shit ton of tabs opened in your browser?
yeah
like 20 rn
tribal pivot
ohh
low shard
what browser are you using btw?
tribal pivot
opera gx
low shard
opera gx
that browser could led you to have trouble, many users reported to have issues with it for using wokada
also, it eats even more ram for it's fancy effects
which is why we don't suggest it
tribal pivot
o
low shard
all those fancy things eat performance ofcourse
tribal pivot
what browser is good
low shard
what browser is good
usually chrome or firefox don't give issues with it
you could still use operagx, but just be aware of more ram being eaten and possible issues
tribal pivot
ok ill use chrome
tribal pivot
you could still use operagx, but just be aware of more ram being eaten and possi...
also why does my voices not sound realistic
low shard
also why does my voices not sound realistic
show your wokada settings
tribal pivot
it sounds realistic for like a second but then it goes back to sounding like a voicemod voicechanger
low shard
!give-media-perms 1h @tribal pivot
tribal pivot
!give-media-perms 1h <@1001566588702560256>
hold on wait iam reinstall cuz avast got mad at it and i cant remove it from the thing
ima
low shard
how to know if the w okada is safe
also, the program is open source, the entirety of the code is public to see at https://github.com/deiteris/voice-changer , we even use the programs ourselves
I'm guessing you don't know coding, but just letting you know
tribal pivot
yeah i dont😭
low shard
hold on wait iam reinstall cuz avast got mad at it and i cant remove it from the...
yeah anti virus sometimes could have false positives, it's normal, it's better you make an exception for the program folder
tribal pivot
yeah anti virus sometimes could have false positives, it's normal, it's better y...
yeah especially avast
low shard
yeah i dont😭
TL;DR: the code is public, meaning we can check all the lines and all the things the program does, and we wouldn't suggest or use programs that we know have malware
that's what open source is, public code
low shard
yeah especially avast
let me know when you're done and got in wokada
low shard
alr
low shard
uh now its saying this
show the command line window
and also show the wokada settings
what's your pc gpu and also show the wokada settings
f0: rmvpe without onnx
extra: 2.7
be sure windows and your gpu drivers are up to date
tribal pivot
f0: rmvpe without onnx
extra: 2.7
be sure windows and your gpu drivers are up t...
my drivers are up to date
low shard
my drivers are up to date
what about your windows version?
tribal pivot
what about your windows version?
windows 10 and updated my windows this morning
low shard
windows 10 and updated my windows this morning
could you try to:
- download vac lite if you haven't (3rd step of the guide)
- restart your pc and the program
- set the output in wokada to line 1
tribal pivot
could you try to:
1. download vac lite if you haven't (3rd step of the guide)
2....
ok hold on
tribal pivot
could you try to:
1. download vac lite if you haven't (3rd step of the guide)
2....
itt works now ty :D
low shard
itt works now ty :D
yw, about the chunk:
it controls the delay, it's better you find your own best chunk by closing all useless programs in background, leaving only wokada, the browser and the program you want to use open, click start, and check the perf value at the top left of wokada, then set the chunk to that value + 60 (or generally just a bit higher)
also, there is a force fp32 setting in the advanced settings that will make your model a bit higher quality and more stable, at the cost of some delay though
tribal pivot
yw, about the chunk:
it controls the delay, it's better you find your own best c...
thank you :D
low shard
thank you :D
you're welcome, for any future issues ask there or make your own post #1192011222023950368
tribal pivot
ok :D
kindred pine
any one know of any good ai software's? like am using Okada, but the latency is so unbearable long
main crag
guys how do i use model without WebUI in python or js?
and i just can't find config for shape
low shard
any one know of any good ai software's? like am using Okada, but the latency is ...
be sure to be using wokada deiteris fork and not youtube tuts, video tutorials are all old
share a screenshot of your wokada settings
!give-media-perms 1h @kindred pine
low shard
guys how do i use model without WebUI in python or js?
so use rvc for pre-recorded audios in python/js?
main crag
so use rvc for pre-recorded audios in python/js?
i keep getting those errors
RuntimeError: mat1 and mat2 shapes cannot be multiplied (80x77 and 768x192)```maybe try:
- https://pypi.org/project/infer-rvc-python/#history
- https://github.com/blaisewf/rvc-cli
- use as an api button on the bottom of the web ui
kindred pine
!give-media-perms 1h <@470234117355733003>
low shard
damn that's ancient
kindred pine
XD
low shard
you deffo used a youtube tutorial
the one you got has performance issues, it's an old version of the original wokada
plus vb cable gives random issues on windows
delete everything you got off youtube
-realtime
low shard
you deffo used a youtube tutorial
u deffo did since the settings were even wrong too
only written guides are updated
kindred pine
so its not the right software?
low shard
so its not the right software?
exactly, the one you got is an old version of the original wokada
Wokada has 2 main versions:
- Original made by Wok
- Deiteris fork (modified version) made by Deiteris
each version has it's own updates
the latest deiteris fork is better than the one you're using
kindred pine
possible to send me the link to the actuall software?
low shard
possible to send me the link to the actuall software?
i gave you the guide
low shard
^
the 1st link
you have to read the guide to understand how to download it and how to use it
kindred pine
oh ok ok, am high af so kinda just spacing out XD, thx ill be reading the guid then
low shard
oh ok ok, am high af so kinda just spacing out XD, thx ill be reading the guid t...
alright, let me know
pure gust
I'm stuck because a voice is downloading but it's making me spit because it's bugged help me please..
low shard
I'm stuck because a voice is downloading but it's making me spit because it's bu...
elaborate more
- what's your pc gpu
- what tutorial link are you using
- a screenshot of the program settings
- what you want to do
- what model link did you use
be sure to not use youtube tutorials
pure gust
Ok
low shard
video tutorials are all old, only updated ones are the written ones
pure gust
I have photos of what it does to me
low shard
I have photos of what it does to me
you need to provide all the informations I said
!give-media-perms 1h @pure gust
low shard
- what's your pc gpu
- what tutorial link are you using
- a screenshot of the pr...
be sure to provide everything I said here
@pure gust Don't post in #🏙│ai-images
pure gust
Ok
low shard
that's not the right channel for doing that, I deleted your messages
this is the right support channel
pure gust
Ok
low shard
Ok
don't use voice.ai, it's paywalled and sucks
you still need to provide all the info
low shard
- what's your pc gpu
- what tutorial link are you using
- a screenshot of the pr...
^
reply to all those questions
if you don't, we can't help without any info, AI isn't a program that can run locally on a 10 year old pc, we need to check if your pc is good enough
also be sure to not follow video tutorials
low shard
Ok
you have image permissions here, use this channel not #🏙│ai-images for help
pure gust
you need to provide all the informations I said
the processor is an ADM Ryzen 53500U with Radeon Vega Mobile Gfx
what is a tutorial uh ..
low shard
the processor is an ADM Ryzen 53500U with Radeon Vega Mobile Gfx
that's the processor, aka CPU, Central Processing Unit
I asked for your PC GPU, does your pc have a dedicated gpu?
You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU
the gpu, graphics processing unit, is very important for intensive tasks like graphic, gaming, and AI
pure gust
AMD Radeon(TM) Vega 8 Graphics
low shard
AMD Radeon(TM) Vega 8 Graphics
that's integrated graphics, which is weak and not good for intensive tasks like AI
are there any other GPUs?
pure gust
Uh...
low shard
Uh...
if not, your pc is cooked since it's too weak to run ai locally
locally means that the program will run on your hardware
cloud means it will run on a remote good pc
it's better you use cloud
the only working cloud method is Wokada Deiteris Fork
wokada is a program designed to run RVC mdoels (speech to speech) in realtime for calls/games, and deiteris fork is a modified version which is better
kaggle is a cloud provider by google, it gives you 30 hours weekly of free gpu, but it's hard to use and needs a phone number
you don't have much other choices than this, voice.ai makes you pay and also stays always on your pc to use the pc power for their own services
pure gust
I see
pure gust
follow that thing I told you and let me know
I succeeded I think
low shard
I succeeded I think
are there any issues
u can also show a screenshot of the wokada so I can suggest you settings
pure gust
are there any issues
Actually no, there are mistakes uh
low shard
Actually no, there are mistakes uh
elaborate the issues
!give-media-perms 1h @pure gust
you can also send screenshots
pure gust
I managed to do it but it gives errors at the end because I don't really know how to do it
restive kiln
-realtime
pure gust
actually it works omg
low shard
I managed to do it but it gives errors at the end because I don't really know ho...
that happened because you probably runned it twice, opening multiple tunnels
low shard
actually it works omg
if u want I can check your settings
pure gust
that happened because you probably runned it twice, opening multiple tunnels
Mmh
probably
low shard
im looking for the rvc channel
all help channels got merged into one
so it's easier for users to find help, and not to post the same question in 6 different channels
verbal mural
help bro
low shard
please elaborate
low shard
probably
yeah, also you need a virtual audio cable to make the wokada work in your discord vc/game
follow only this part of this guide https://rentry.co/forkvoicechangerguide#virtual-audio-cable for getting vac lite
verbal mural
i need a good free ai to create animated videos for tiktok with my script
low shard
Ok
also after that, make sure to set your wokada output to line 1 and the other program (like discord vc) input as line 1
you can also send a screenshot of your wokada settings
low shard
i need a good free ai to create animated videos for tiktok with my script
so, text to video?
text/image to video AIs:
- Locally (runs on ur pc):
- pyramid flow (Image/Text to Video)
- cogvideox 1.5 5b: Image to Video, Text to Video
- Cloud (remote good pc, running on an online website for example, easier to setup):
- Weights.gg
- pyramid flow (Image/Text to Video) (HuggingFace Space)
- OpenAI Sora (paid only, in some countries)
- lumalabs
- Hailoua AI
verbal mural
thhank you bro i will try them
hallow thistle
im looking for the rvc channel
Help channels for RVC, W-Okada, AI image and others all merged into one, but there's also #1192011222023950368.
restive kiln
Ohhhhhhhhhhhhhhh, btw is the colab version abandoned now or just stored somewhere
last time I used it, it was broken
pure gust
also after that, make sure to set your wokada output to line 1 and the other pro...
It's good ?
low shard
Ohhhhhhhhhhhhhhh, btw is the colab version abandoned now or just stored somewher...
what colab version? what link are you talking about? what you want to do? what's your pc gpu?
low shard
It's good ?
set your usual devices as default ones
slate viper
what version i download on hugging face can someone send me link
pure gust
set your usual devices as default ones
Ok
low shard
what version i download on hugging face can someone send me link
elaborate
what do you want to do?
what's your pc gpu?
slate viper
what's your pc gpu?
3060 rtx and intel 7
low shard
there's thousands of AI programs and models, with different versions for each gpu and OS
low shard
3060 rtx and intel 7
nice but what do you want to do?
hallow thistle
It's good ?
The green tick is supposed to be your main speaker/microphone.
slate viper
nice but what do you want to do?
can you send me download link for the ai voicechanger
low shard
can you send me download link for the ai voicechanger
realtime for calls/games right?
if so, wokada deiteris fork
slate viper
realtime for calls/games right?
yes
low shard
-realtime
slate viper
apri it bro thank you
low shard
apri it bro thank you
yw, let me know for any issues
hallow thistle
3060 rtx and intel 7
Intel 7? Is it the Intel Core CPU generation or Intel Core Ultra? 
slate viper
Intel 7? Is it the Intel Core CPU generation or Intel Core Ultra? <:iorithink:10...
intel(R) Core(TM) i7-10700F CPU @ 2.90GHz 2.90 GHz
my pcs in shambles need to get a new one fr
hallow thistle
intel(R) Core(TM) i7-10700F CPU @ 2.90GHz 2.90 GHz
That Intel CPU is at least decent enough to run AI.
verbal mural
text/image to video AIs:
- Locally (runs on ur pc):
- [pyramid flow (Image/Te...
nothing works bro and the most are paid
low shard
nothing works bro and the most are paid
elaborate more the issue, what doesn't work? what did you try? what's your pc gpu?
also, if you're using cloud, you can't expect free service 24/7, those run on gpus that are extremely expensive
verbal mural
i paste my script, that should create a 1minute tiktok but i only get 1 second clips or videos with stock images
hallow thistle
nothing works bro and the most are paid
Then there's nothing you can do if you think paid services are bad and you don't have a great PC with great GPU to run AI video generator locally. 
verbal mural
my pc is not a problem
low shard
i paste my script, that should create a 1minute tiktok but i only get 1 second c...
those are tools for text/image to video, not auto tiktok generators with family guy on top and a random subway surfer gameplay below if that's what you're looking for ?
verbal mural
i just have no money to buy ai
hallow thistle
Still.
verbal mural
those are tools for text/image to video, not auto tiktok generators with family ...
thats not what i want either
low shard
thats not what i want either
can you show an example then?
verbal mural
i dm you my script ok?
hallow thistle
low shard
i dm you my script ok?
it's better we keep things here
you can just show an example of a video that gave you inspiration, or one you want to do similarly
hallow thistle
Do not direct message someone without their consent.
hallow thistle
Crazy.
low shard
i got this idea without any inspo
can't you just explain what you want exactly?
like an ai that automatically generates the video you want, edits it, adds subtitles and automatically posts it for content farm ?
hallow thistle
At least just say some keywords of it, it ain't that hard.
It ain't hard to tell.
verbal mural
english is bad cant explain
hallow thistle
are you kidding
No shit.
latent cypress
https://tenor.com/view/tung-tungtung-tungtungtung-sahur-tungtungtungsahur-tungtu...
TUNG TUNG TUNG SAHOOR!!!
hallow thistle
That is whan I can think of a video, according to what you say.
verbal mural
tell me what ai to use please
hallow thistle
If you don't explain some more, like voice acting, TTS or image/video generator, then I don't know what else you're looking for.
verbal mural
this animation style is fine
verbal mural
bro i just cant talk in english
this gay ass comunity all sitting on dick
no one with real brain
latent cypress
english is bad cant explain
you can use chatgpt to translate your message, tell gpt to fix the grammar and make it sound natural
that's what i always do
verbal mural
true but this crazy i just improve like this
hallow thistle
If you can't explain things right now, just get out from here for a while and come back when you know you can.
latent cypress
no one with real brain
the problem is you lol, don't blame anyone here
verbal mural
i did explain but no one get it bruh like is it hard to understand i just need invideo for free
i dont even know good paid ai
hallow thistle
No one can read what's going on your mind, so you'd have to be more precise about what you're gonna do I guess.
verbal mural
this shit crazy yall are maxed out nerds
latent cypress
i did explain but no one get it bruh like is it hard to understand i just need i...
hallow thistle
Don't try to play a victim like that. 
verbal mural
thank bro i will try
hallow thistle
On Weights, there's a feature that you can generate video there, but you'll have to pay for premium to use it.
latent cypress
thank bro i will try
https://youtu.be/c38vtLw1nSk?si=cG2s9-m07_k7Mysy here's the tutorial, idk if this tutorial is outdated or not for now
verbal mural
bro nice ty
hallow thistle
There's Luma Dream Machine. I did these videos last year. While you can use the service for free for a limited period, you may often encounter the long queue number.
Only if I could explain at least understandable to non-English speakers while my English is too good though.
hallow thistle
You can generate some videos there for free, but again you will encounter queued task there if you have free plan. That's pretty much it. For subscription plans, you will have to go read everything by yourself. Same goes to some other websites.
knotty moth
is video also free
it is more demanding and there's nothing more affordable than queue for several hours
hallow thistle
I don't know how would you even understand this, but this is my best I could explain.
brittle wing
i paste my script, that should create a 1minute tiktok but i only get 1 second c...
@redwonder3 who are you 🤣
tough fiber
is it possible to use one index file on different models , on voice changer?
odd shale
is it possible to use one index file on different models , on voice changer?
Well, normally it's not necessary to use index on realtime, but i think you only should use it on it's respective model.
slim bay
I'm looking for a AI model which have better and high accuracy of STT, TTS, emotion deduction, video analyzing and generating the perfect report's. as per my research i want not able to find any multimodal which have all those thigs, like deep gram, whisper, hume, proctring so on
hallow thistle
i want diddy ai
For a voice model, look him up in #1175430844685484042. 
hallow thistle
is it possible to use one index file on different models , on voice changer?
It's possible to use index from one voice model with another, I tried it on RVC and Applio, but as an experiment. While using index on W-Okada is possible, the program gonna use more resource than setting it to none.
violet rose
what is the best into 50 & 100 eposh
ancient ocean
I am traning Voice Model for singing with 1:10 hours long high quality dataset and TITAN Pretrains. Should i stop training or train futher? It already trained to 650 epochs. In tests by ear i can't hear significant diference between 330 and 650 epochs
potent cryptBOT
VoiceMaster+
simple ore
I am traning Voice Model for singing with 1:10 hours long high quality dataset a...
well, I can tell you that if you take klmn 4.9 and cut your dataset down somewhat, you may get better result that with titan
titan is a "finetuned" pretrain, not sure how big the variety is there, but it is not singing pretrain
ancient ocean
well, I can tell you that if you take klmn 4.9 and cut your dataset down somewha...
I used the single WAV file by joining all samples. Normalized and cleaned. I suppose the Preprocess step does it, doesn't it? What is "klmn"? Which pretrain can be ideal? I want at least 40k sample rate
left wadi
@low shard
left wadi
holy shit that's ancient
whats ancient? LMAO
low shard
whats ancient? LMAO
your program
left wadi
your program
oH
left wadi
yeah
fading barn
If I want to use w-okada on server, is there anything I can use to block out the noise from me typing from being picked up?
low shard
-realtime
left wadi
Me or him?
low shard
Me or him?
you
low shard
am I gonna have to read the whole thing?
yup
it's not that long
you have to follow only the parts related to windows nvidia, since you got an nvidia gpu and your OS is probably windows
low shard
If I want to use w-okada on server, is there anything I can use to block out the...
I don't think so
maybe @pastel oak knows another way, but server shouldn't be able to use sup2, sup1, nor echo suppression
fading barn
I've seen some people say nvidia broadcast when searching here but I'm not sure how to go about setting it up, trying to figure it out
yeah server can't use any of those
left wadi
I already have nvidia installed its essential
ancient ocean
klm4.9 is a pretrain
Is it trained with Russian vocals? And how about cutting the dataset? How should I cut the dataset? Should I use 48k sample rate?
low shard
I already have nvidia installed its essential
with windows nvidia, I meant the wokada deiteris fork windows nvidia version
everything you got off youtube tut was really old
left wadi
with windows nvidia, I meant the wokada deiteris fork windows nvidia version
oh you meant that
low shard
oh you meant that
and if you downloaded vb cable, uninstall it
it gives random issues on windows
youtube tutorials get so easily outdated
left wadi
idk what that is but I dont think that I installed it thankfully
low shard
which is why we usually don't make them, written guides are way easier to keep updated
low shard
idk what that is but I dont think that I installed it thankfully
did the yt tut make you install any other thing along the wokada
fading barn
Ah I figured it out for server noise reduction, you put your regular mic input into broadcast, enable the sound settings then set the input for okada to nvidia broadcast
left wadi
I dontt think that I have this installed do I need this?
fading barn
do you have an nvidia gpu?
left wadi
do you have an nvidia gpu?
yeah
fading barn
then no
low shard
I dontt think that I have this installed do I need this?
no you don't, you need only to install the wokada deiteris fork nvidia windows version
fading barn
use th nvidia
left wadi
perfect
low shard
ofc not the 5000 serie, since you got a 3000 rtx serie
also delete the older version of the original wokada that you got
left wadi
no you don't, you need only to install the wokada deiteris fork nvidia windows v...
yeah couldnt find the link to that
low shard
and also make sure you don't skip the 3rd step
left wadi
just delete the folder
oh alright bet
fading barn
yeah couldnt find the link to that
Directly above the picture you sent is a link for the 5000 series nvidia, and above that is for the normal nvidia gpus
prisma dove
Can smb help me with Voice changer?
low shard
yeah couldnt find the link to that
low shard
Can smb help me with Voice changer?
!howtoask
low shard
please elaborate
- what's your pc gpu?
- what issue are you facing?
- what tutorial link did you use?
- show a screenshot of your program
left wadi
Directly above the picture you sent is a link for the 5000 series nvidia, and ab...
found it thanks
low shard
!give-media-perms 1h @patent trellis
dull ironBOT
!give-media-perms 1h <@678344927997853742>
low shard
oops wrong user lmao
!give-media-perms 1h @prisma dove
also, be sure to never follow youtube tutorials for wokada, they are old
prisma dove
- what's your pc gpu?
- what issue are you facing?
- what tutorial link did you ...
To be more accurate,
I have laptop with rtx3050 , and Iris xe
Issue was that i choose in vc Rtx , but it not using at all.
Tutorial was my frendo
left wadi
and also make sure you don't skip the 3rd step
this is the third step correct?
low shard
this is the third step correct?
yes, get the vac lite
left wadi
this guide is a bit confusing sorry
low shard
To be more accurate,
I have laptop with rtx3050 , and Iris xe
Issue was that i...
show a screenshot of your wokada
prisma dove
show a screenshot of your wokada
low shard
also thanks for the info
low shard
yeah your friendo is cooked too then
that is an old version of the original wokada
it's like over a year old
prisma dove
yeah your friendo is cooked too then
but he has nothing problemo at all
low shard
your friend probably used a youtube tutorial
low shard
but he has nothing problemo at all
the issue is your settings, but it's not worth to fix it in that program, because the version you and your friend got have worse performance and worse quality
both of yall should first uninstall this version
then fix the settings
prisma dove
oke
low shard
@prisma dove please delete the version you got, then read the guide for the wokada deiteris fork https://rentry.co/forkvoicechangerguide
also did your friend make you install something called like vb audio cable?
low shard
can you send some link of rvc? please
this isn't rvc, this is wokada,
RVC and Wokada are 2 different programs
RVC = Retrieval-based-Voice-Conversion, the best Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models
Wokada = uses RVC for realtime inference
You need wokada since you will use it in vc
Wokada has 2 main versions:
- Original made by Wok
- Deiteris fork (modified version) made by Deiteris
each version has it's own updates
the version you're using is an old one of the original wokada
which has performance and quality issues, so don't use it
I already gave you the link for the wokada deiteris fork tutorial, https://rentry.co/forkvoicechangerguide
you need to read it
there's no updated video tutorial
prisma dove
oke
low shard
why that reaction? written guides don't get outdated like video tutorials do
we can freely keep written guides up to date everytime
unlike video tutorials that are harder to keep up to date since they require more time
I hope you understand
prisma dove
ofc, just cause im lazy asf
low shard
for any issues or question you can ofcourse ask here
prisma dove
ty for support
low shard
ofc, just cause im lazy asf
we know that users find video tutorials easier, but making a video tutorial that would comprehend every single os with every single gpu brand would be difficult, along with the fact that AI progresses at a sonic speed
low shard
ty for support
you're welcome, we are here for that
prisma dove
we know that users find video tutorials easier, but making a video tutorial that...
nono, i think that's tutorial link is more easier than i uses later for another programs. And reason is more correct that could be
low shard
nono, i think that's tutorial link is more easier than i uses later for another ...
Goodluck and let me know how it goes for you then
low shard
this guide is a bit confusing sorry
are you facing any issues?
low shard
what am i doing wrong
rvc mangio is ancient, that's what you're doing wrong
echo turtle
ahh alright
low shard
mangio rvc fork has been abandoned since 2023
low shard
NEVER follow youtube tutorials for rvc
echo turtle
lmao sorry
low shard
lmao sorry
nah don't say sorry lol
low shard
what should i use
what's your pc gpu? and what do you want to do?
echo turtle
got a 4070 laptop gpu
and im just using voice models to sing songs
like homer simpson sings not like us
etc.
echo turtle
thank you!
low shard
thank you!
cloud (remote good pc) also exists, which is easier but there are free tier limits (like time) and are less reliable, if you want I could tell you them but it's not worth it when your laptop is good for local inference (inference is using the models basically)
low shard
but thank you for the advice
you're welcome, let me know for any issues
echo turtle
i will
simple ore
Is it trained with Russian vocals? And how about cutting the dataset? How should...
no, there's no russian vocal, but there are high pitch samples that help with high pitch inference on a finetuned model.. one big file is fine, dont use process effects in applio if you already has it normalized and cleaned, do audio cutting in preprocess
ancient ocean
no, there's no russian vocal, but there are high pitch samples that help with hi...
OK. Thank you very much 🙏🏻. How about 40k and 48k . Should I try 48k ?
simple ore
depends on the dataset, check with spek
pastel oak
I've seen some people say nvidia broadcast when searching here but I'm not sure ...
Server mode cant use those, need 3rd party program
fading barn
I figured out how to use broadcast with server mode 🙂
river cairn
Hello, I'm currently using an old version of w-okada's mainline RVC for real-time voice changing. It's been a few years since I checked the space, and it seems like there have been many advancements. Are the w-okada mainline RVC and the Deiteris fork currently the only/best real-time voice changers, or are there other/better options?
echo turtle
you're welcome, let me know for any issues
i keep getting an error now when i hit convert i am using applio
prisma dove
Goodluck and let me know how it goes for you then
I couldnt find in guide the settings blank??
prisma dove
I couldnt find in guide the settings blank??
Is there easier use client with low-speed internet, or server?
nocturne stone
whenever i try using server it doesn't work but when i go to client it does work, can anyone help me?
low shard
i keep getting an error now when i hit convert i am using applio
show a screenshot
!give-media-perms 1h @echo turtle
echo turtle
low shard
I couldnt find in guide the settings blank??
wdym?
what browser are you using and show a screenshot
low shard
whenever i try using server it doesn't work but when i go to client it does work...
show a screenshot and elaborate the issue
!give-media-perms 1h @nocturne stone
low shard
show a screenshot of the CMD, and also a link to the model you're using
nocturne stone
show a screenshot and elaborate the issue
so when i use server my voice doesn't work, and when i use client it does with the same output device and input.
low shard
operagx
operagx is known for giving issues for wokada, also it eats more ram so more performance because of its fancy effects
we suggest chrome or firefox
prisma dove
ok
prisma dove
no