✨│ai-help

[edgy quiver]

Is vonvox or the other fork of okada better

[viral mason]

Vonovox

I'll get you the downloads

[finite wind]

as nvidia user idk about vonovox

[viral mason]

It's currently the best

[finite wind]

it gives poor results than wokoda fork

for me at least

[edgy quiver]

Okok no harm in trying ig

[finite wind]

everyone says it's the best but it was opposite for me

[viral mason]

https://huggingface.co/dr87/vonovox/resolve/c8034f5f6d50648a8109bb4f847182362e2b779b/Vonovox_beta_17_11.zip

https://software.muzychenko.net/freeware/vac470lite.zip

[low shard]

they have different pros and cons

it would be best you read the guide to get more info

[viral mason]

if you dislike the results of how this one sounds I will get you the correct download for Tg fork

[low shard]

you can simply try both

[edgy quiver]

Okok

[finite wind]

but besides vonovox is able to handle 50s serise nvidia and better at using nvidia graphics card

what's the other reasons why vonovox is better

[edgy quiver]

Are both links to vonvox?

[finite wind]

because the very same model I tried on both, vonovox clearly gave poor results

[edgy quiver]

Also thanks for all the help i really appreciate it

[hardy yew]

improved RMVPE, improved feature extraction, better volume handling, WSOLA implemented for better stitching

in general vonovox was made from scratch and nicely optimised for realtime RVC

so it's expected to be much better and more stable than w-okada

[viral mason]

the first one if Vonovox the second is a virtual audio cable that will connect your voice changer to discord or games

[finite wind]

all that and the model I was using on wokoda fork didn't work at all on vonovox for some reason

as if it wasn't utilizing my GPU or the model at all

[edgy quiver]

Ok gotchu

[finite wind]

it wasn't embedder or RMVPE issue too which made me super confusing

[hardy yew]

well yeah i don't guarantee that it's always better and works the best for everyone, just that it's expected to be better due to tons of improvements

[analog obsidian]

embedder = language
rmvpe = pitch tracking

[edgy quiver]

what about the deiteris fork

is it any good

[hardy yew]

better to use tg instead i think

tg is a fork of deiteris after all

[viral mason]

test for both using the same model

[finite wind]

hmmmmmmmmmmm

[viral mason]

ignore the fl studio setup since my setup is very different from a normal user

I use it to add effects ect

[low shard]

This is a General AI Discord Server and there are many voice changers, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, AI Covers, E Girl Trolling / Catfishing or Roleplay
• the tutorial link used

[low shard]

wokada deiteris fork hasn't been updated in a while, tg-develop continues off it

[viral mason]

ew

[edgy quiver]

right, so if i understood correctly, vonvox is better for nvidia gpus compared to the tg-develop fork

[low shard]

smh

[edgy quiver]

we all start somewhere T-T

[viral mason]

technically yes but it depends on what you want

you're being sent to the shrimp dimension

[hardy yew]

https://tenor.com/view/kaetha-fly-nope-eject-rocket-launch-gif-17169750

[toxic perch]

https://tenor.com/view/funny-shrimp-dance-bumping-gif-16904084

[viral mason]

any who want to use "girl voice changer" shall be punished for their crimes

[edgy quiver]

lol

[toxic perch]

fr fr

[hardy yew]

https://tenor.com/view/egirl-luiguy-andreegirl-girl-gif-18978519

[viral mason]

unless the person is trans and wants to hide their male voice they will be sent to hell for being yucky

[edgy quiver]

what model did u use here its sounds pretty good

[viral mason]

my own model of Jax from the amazing digital circus, it's not public at least here due to me not wanting to get murdered by the voice actor

[edgy quiver]

lol okok makes sense

[viral mason]

yah

[edgy quiver]

do u know any public male models i could use, all the ones in the voice models thing are the trolling deep eboy ones

[viral mason]

just look for a character from a show or smth you like the sound of

https://discord.com/channels/1159260121998827560/1175430844685484042

I make a lot of models of random game and show characters

[viral mason]

from that quick test I did tho what did you think @finite wind and @edgy quiver

which sounds better?

just delete the folder it's in

since it isn't a proper app that will remove it completely

[low shard]

do you want to remove it or just fix it?

[edgy quiver]

lowkey, they kinda sound the same, vonovox does seem a bit better tho (it sounds more full)

[finite wind]

more smooth pitch transistioning

but somehow my pc HATES vonovox

so I can't use it

[edgy quiver]

yeah that

[viral mason]

what gpu do u have?

[finite wind]

3060ti

so in paper, it should be no problem at all

[viral mason]

it should work fine, have you tried the beta?

[finite wind]

I was expecting an improvement but here I am

yep same version as yours

actually tried stable version too

[viral mason]

how odd

[finite wind]

jittering, lagging, GUI window glitching etc

so I'm staying away

[viral mason]

how odd

[finite wind]

monitored GPU, CPU, RAM usage and it was using same as wokoda fork

so it wasn't leaking for sure

or not even using in the first place

[viral mason]

you could join the official vonovox discord and have a smart person (not me) look into it

[finite wind]

maybe in another day

[viral mason]

👍

[hardy yew]

CUDA

[solemn swan]

does anybody know why kie ai says The Seedance 2 API is not yet available. Please stay tuned and be patient.

[viral mason]

don't use yt tutorials they're outdated, what are u using the voice changer for btw just curious

[edgy quiver]

[viral mason]

that guy is a weeeeeirdooo

[edgy quiver]

i love the emojis in this server

[viral mason]

see why there is a hall of shame channel

you got a live example of those weird ppl that join

[edgy quiver]

peak entertainment tho

[viral mason]

I think that guy should just come out and be gay if he wants to date his friend

nothin wrong with that

[hazy dagger]

yoo i need a model rec.

i dont know how to find the most suitable one

[low shard]

a model rec?

[hazy dagger]

a suggestion basically

which one i should use

i am sorry for the silly question

[low shard]

for what? there are thousands of models like ben10 etc

are you trying to do e girl trolling / catfishing?

[hazy dagger]

umm a deep male voice

[low shard]

or e boy trolling / catfishing?

[hazy dagger]

i am not trolling anyone 😭

i just need a deep voice

thats it

[viral mason]

thank you for being normal

[hazy dagger]

lol

[finite wind]

so uh

should I export it as 44k or 32k

[viral mason]

I always train 32k so idk

[hardy yew]

yeah, just go for 32k

[finite wind]

sure why not

[hardy yew]

i mean, applio will resample it if needed

but less work for it if you already feed it with appropriate sample rate data

[finite wind]

yeah it was my plan until this god damn mp3 decided to be 19k

it's like I'm seeing a forbidden A4 format

[viral mason]

just look for a character or person with a deep voice here https://discord.com/channels/1159260121998827560/1175430844685484042

[finite wind]

wait worm I got a question for keggle thing

do I upload custom pretrain just like I would upload a dataset?

or is there any other way

[hardy yew]

sure, you can do that

the most simple way i guess

[finite wind]

oh okay

[hazy dagger]

how do i sort it based on popularity?

[viral mason]

not really a way to sadly just look for the newest version of whoever you want to try

[hazy dagger]

oh okay mrow

[viral mason]

@finite wind

this is how

[finite wind]

ah

I already uploaded a bunch of pretrains from my drive

[viral mason]

mrow indeed

[hazy dagger]

which one is your favourite mrow?

[viral mason]

the models I make on my own :3

[hazy dagger]

can i get one of those

[viral mason]

why u blushing 😭

[hazy dagger]

(⁠〒⁠﹏⁠〒⁠)

[edgy quiver]

can a man not be embarrassed

[viral mason]

he can, I just don't see why he is

[hazy dagger]

Not everything needs a reason mrow (⁠〒⁠﹏⁠〒⁠)

[viral mason]

https://cdn.discordapp.com/attachments/805441044711079936/1311872879725379644/kcv74cvw32yd1.gif

[hazy dagger]

😭 can i just get the model i need to disappear quickly

[viral mason]

idk what model to really give u since I mostly don't make regular person voices

jsut use this one for now, not sure how good it is since it's oldish https://huggingface.co/YourLocalWorm/TADCshid/resolve/main/KingerTADCV1_570e_16530s.zip?download=true

[stark tundra]

what voice models do u make?

[viral mason]

video game characters and show/movie characters

[hazy dagger]

thanks mrow (⁠☆⁠▽⁠☆⁠)

[viral mason]

you're welcome :3

[stark tundra]

nice 😮

[viral mason]

https://tenor.com/view/bunny-watching-baby-sensory-baby-sensory-gif-5238469316282201243

[novel onyx]

Hi Does anyone know where inference can be done? I was recommended Applio but I'm getting an error.

Everything was fine, but today it won't let me in.

Does the same thing happen to you?

[finite wind]

ma I'm terrified

[hardy yew]

paid for the entire CPU

need to use the entire CPU

[hazy dagger]

hey, these hugging face links are safe, right ?
for example: https://huggingface.co/Homiebear/JinwooSung_490e_61250s/resolve/main/JinwooSung_490e_61250s.zip?download=true

i got it from here #1488328191428001912

[hardy yew]

links are safe, the website is verified

but the files you download can be virtually anything

[hazy dagger]

its a zip file with .index and .pth in it

[hardy yew]

yeah, that's expected

no one really makes effort to put malicious stuff there

but that's what you want to download anyway, a .pth model and optionally an .index file for it

[robust cloak]

Are you actually using GoHighLevel…
Or just paying for it every month?

Be honest.... Are your leads getting followed up instantly…
Or hours later when the energy is already gone?

Do you even have a system…
Or are you still doing things manually and hoping it works?
I work with GoHighLevel setups that look “fine” on the surface…

But behind the scenes?

Leads are leaking.
Follow-ups are inconsistent.
And money is quietly being lost every single day.

What I focus on is simple:
Fix the gaps → increase conversions → drive actual revenue.

Not more tools.
Not more complexity.
Just systems that turn existing leads into booked calls and paying clients.

So let me ask you this...
If someone secretly audited your GoHighLevel setup today…

What do you think they’d find broken?

[finite wind]

I'm going to train a model with just a 2 mintues and 30 seconds dataset

not a single second wasted in this audio file

[viral mason]

yep

[hazy dagger]

@hardy yewThanks for confirming mrow, I’m gonna use it anyway.

[silver lynx]

heyyyyy best friends !

[hazy dagger]

⁄⁠(⁠⁄⁠ ⁠⁄⁠•⁠⁄⁠-⁠⁄⁠•⁠⁄⁠ ⁠⁄⁠)⁠⁄

[finite wind]

pffff

5 steps

[silver lynx]

my voice changer in on my gpu but for some reason its using 50% cpu and 6% gpu

anyone know how ??

[viral mason]

you're very silly

[viral mason]

batch 4 trust

in kaggle it does a double batch size so choose 2

[finite wind]

hey I already chose 4

why you do this to me

[viral mason]

oop

so if you haven't changed the gpu thing it's technically training at batch 8 since it doubles the batch size

[wind thunder]

whenever i use a custom voice model insted of the basic 4 ones my res ms skyrockets does anyone know why

[viral mason]

basic 4 ones?

[wind thunder]

well yk the japanese ones

[viral mason]

you're using an outdated program, what is your gpu (Nvidia or AMD) and what are you trying to do with it, just curious

[wind thunder]

i have an amd gpu

[finite wind]

which one do you use, T4 x2 or a single P100?

[wind thunder]

not even trying anything just tryna talk and ms skyrockets

[viral mason]

1 sec I'll get the downloads you need

[wind thunder]

aight aight

[silver lynx]

the amd one works so good except for my previous mentioned issue

[hardy yew]

i've always just used the P100, actually dunno how it performs with T4 x2

guess it could use a direct comparison

[finite wind]

eh

[viral mason]

https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-dml.zip

https://software.muzychenko.net/freeware/vac470lite.zip

[wind thunder]

tysm

[hazy dagger]

how much do i set the pitch btw?? (⁠☆⁠▽⁠☆⁠)

[viral mason]

for vac lite (the virtual audio cable) run setup64 and install driver, for the voice changer run mmvcserversio

[viral mason]

since it's a male voice just use pitch at 0

u can play around with it

[finite wind]

it finally broke after 290 epochs

[viral mason]

broke?

[hardy yew]

bork

[finite wind]

2minutes and 30seconds dataset

I'd say more than I expected outcome from this small dataset

unusable still

[viral mason]

I've made plenty of small dataset models that are usable

[finite wind]

I tested 290 epochs and 150 but I guess 4 batch did get to me I guess

which epoch did you settle on as a small dataset model?

[viral mason]

4 for most but 2 if it's really small

remember the double batch size thing in Kaggle

[hardy yew]

double batch if using double GPU

for single GPU it's normal, no?

[viral mason]

idk, just for kaggle specifically

it does that

[finite wind]

batch 4 on 2 gpus and it was batch 8 basically

[hardy yew]

if you had 2 GPUs locally the same thing would apply

[viral mason]

yea

[hardy yew]

batch size doubling

it's not a kaggle thing, i think

just a multi-GPU

[finite wind]

anyway

[hardy yew]

the batch size is per GPU

[finite wind]

gotta try tomorrow

[viral mason]

colab and local I believe don't do it tho

[hardy yew]

is colab multi-GPU though?

[edgy quiver]

I had another question which one is better vonvox beta or the stable vers

[viral mason]

beta is much better

[finite wind]

testing features in them right?

usually in any software field betas are unstable but has newer features but idk about vonovox's case

[edgy quiver]

Ohhhhhh okok

[viral mason]

the beta does have newer features but really isn't unstable at all

[hardy yew]

arguably beta is more stable than the release xd

[finite wind]

why that stable hasn't reach me is

just beyond me

[hazy dagger]

oh ok i am sorry for bothering you again but if you could tell me what to keep index at?

[finite wind]

generally below 40 but depending on a model, it can go up quite a lot

test every 0.10 and see which works the best

[viral mason]

Keep it at 0, it's ok btw u can bother me all you like since I'm lonely most of the day

[hardy yew]

not true, you have us

[viral mason]

Very true

Bother me, I'm lonely

[hazy dagger]

awh don’t say that… I’ll keep you company then 🥺

[ebon tapir]

Help why does this sound like a robot 😭

[median monolith]

After weights shutdown, I uploaded my voice models to huggingface, and uploaded one of them to this site called https://voice-models.com/ . theres just a problem when I tried to make a cover with it (image below):

what does that actually mean?

these are the links to the model btw

[viral mason]

<@&1159293140440723499>

[proven hill]

wow… i must join…

[sand pewter]

People I'm searching for new google collab links yo make my voice model but every video I found is a scam or outdated, my old collab link don't work anymore do you know any LINK with an easy way to make my model? I saw many videos changing the method I used before

[winter oriole]

could AI reverse engineer a product's price to a satisfactorily high degree of accuracy? how would that work in theory? (sry im lk new to all this help)

[astral tangle]

any chance someone could help me understand this step for applio?

im at step 2

got v5.7.1 HIP SDK installed as the documentation says, got the correct version of the archives, just confused on what i need to do with them

[simple ore]

New drivers require a different approach

what's your GPU?

[astral tangle]

Rx 6700 xt, i have a thread in the forums

I verified i have the correct archives and HIP SDK

[simple ore]

new drivers require hip sdk 6.4.2

https://github.com/likelovewant/ROCmLibs-for-gfx1103-AMD780M-APU/releases/download/v0.6.4.2/rocm.gfx1031.for.hip.6.4.2.7z

[astral tangle]

so the documentation is outdated then?

[simple ore]

the zip has a library folder and rocblas.dll

the dll replaces the one in %HIP_PATH%\bin\

the library folder replaces library folder under %HIP_PATH%\bin\rocblas

the rest of the guide

what I said above is step 3.1

[astral tangle]

https://docs.aihub.gg/rvc/local/applio/ so i shouldn't follow this?

in that documentation it says the rx 6700 xt is older drivers/legacy

[simple ore]

@low shard may need to update the guide on docs

hip sdk 6.2.4 would work as well

old 5.7 wont work unless you're using 25.4.1 driver or older

[astral tangle]

of the amd adrenaline software?

[simple ore]

yes

[astral tangle]

aah

[simple ore]

AMD change some internal stuff in 25.5.1 and it broke Zluda

[astral tangle]

so is there an issue if i already isntalled the older HIP SDK version?

like will there be an issue if i now install the new one?

[simple ore]

you can just delete the folder

new one creates a new C:\Program Files\AMD\ROCm\6.4

you can delete 5.7

[astral tangle]

so i just delete the 5.7 one

okay

[astral tangle]

does it matter where i extract applio?

like what should the folder be called=

[simple ore]

C:\Applio

or whatever

just no spaces and no OneDrive

[astral tangle]

so it would be C:\Applio\ApplioV3.6.2?

[simple ore]

just C:\Applio

you dont need another folders inside of it

[astral tangle]

so i should just extract it in the C drive, since it comes in a folder on its own

[simple ore]

the compiled version does not, but if you use winzip/7-zip it jsut creates a folder name same as the .zip

yeah

[astral tangle]

btw sorry for so many questions, im new to this

this should be the system environment variables (path) right C:\Program Files\AMD\ROCm\6.4\bin

[simple ore]

yes

add this as entry to existing Path

[astral tangle]

the commands i should run isnide the applio folder, should i do them 1 by 1 or just copy the whole thing

env\python -m pip uninstall torch torchvision torchaudio
env\python -m pip install torch torchvision torchaudio --upgrade --index-url https://download.pytorch.org/whl/cu118

[simple ore]

1 by 1

open cmd.exe in Applio folder

[astral tangle]

yeah

it says that some stuff will be removed when running the first line

so i assume to pres Y

[simple ore]

yes

[astral tangle]

and now i close cmd and re open to run the next line? its not letting me copy paste the 2nd

nvm

i finished running both lines on the CMD inside the applio folder, but it has given me a warning at the end, WARNING: The scripts torchfrtrace.exe and torchrun.exe are installed in C:\Applio\env\Scripts which is not on PATH. Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location

[simple ore]

that's fine

[astral tangle]

so i'll close it then and finish with the rest of the guide

[simple ore]

you can run the patch .bat from the same cmd

[astral tangle]

by copy pasting this?
rmdir /S /q zluda
curl -s -L https://github.com/lshqqytiger/ZLUDA/releases/download/rel.5e717459179dc272b7d7d23391f0fad66c7459cf/ZLUDA-windows-rocm6-amd64.zip > zluda.zip
tar -xf zluda.zip
del zluda.zip
copy env\Lib\site-packages\torch\lib\nvrtc64_112_0.dll env\Lib\site-packages\torch\lib\nvrtc_cuda.dll /y
copy zluda\cublas.dll env\Lib\site-packages\torch\lib\cublas64_11.dll /y
copy zluda\cusparse.dll env\Lib\site-packages\torch\lib\cusparse64_11.dll /y
copy zluda\nvrtc.dll env\Lib\site-packages\torch\lib\nvrtc64_112_0.dll /y
copy zluda\cufft.dll env\Lib\site-packages\torch\lib\cufft64_10.dll /y
copy zluda\cufftw.dll env\Lib\site-packages\torch\lib\cufftw64_10.dll /y
pause

[simple ore]

just run it

the bat

[coral burrow]

how do i fix this? it keeps telling me to check if the model or the framework is being loaded

[astral tangle]

should i edit the file located at rvc/lib/zluda.py first?

Edit the file located at rvc/lib/zluda.py. Replace the content with the following:

import torch

if torch.cuda.is_available() and torch.cuda.get_device_name().endswith("[ZLUDA]"):
# disabling unsupported cudnn
torch.backends.cudnn.enabled = False
torch.backends.cuda.enable_flash_sdp(False)
torch.backends.cuda.enable_math_sdp(True)
torch.backends.cuda.enable_mem_efficient_sdp(False)

[simple ore]

once you patched

[astral tangle]

okay so then i just run the bat inside the applio folder

C:\Applio>rmdir /S /q zluda
The system could not find the specified archive/folder.

C:\Applio>curl -s -L https://github.com/lshqqytiger/ZLUDA/releases/download/rel.5e717459179dc272b7d7d23391f0fad66c7459cf/ZLUDA-windows-rocm6-amd64.zip 1>zluda.zip

C:\Applio>tar -xf zluda.zip

C:\Applio>del zluda.zip

C:\Applio>copy env\Lib\site-packages\torch\lib\nvrtc64_112_0.dll env\Lib\site-packages\torch\lib\nvrtc_cuda.dll /y
1 archivo(s) copiado(s).

C:\Applio>copy zluda\cublas.dll env\Lib\site-packages\torch\lib\cublas64_11.dll /y
1 archivo(s) copiado(s).

C:\Applio>copy zluda\cusparse.dll env\Lib\site-packages\torch\lib\cusparse64_11.dll /y
1 archivo(s) copiado(s).

C:\Applio>copy zluda\nvrtc.dll env\Lib\site-packages\torch\lib\nvrtc64_112_0.dll /y
1 archivo(s) copiado(s).

C:\Applio>copy zluda\cufft.dll env\Lib\site-packages\torch\lib\cufft64_10.dll /y
1 archivo(s) copiado(s).

C:\Applio>copy zluda\cufftw.dll env\Lib\site-packages\torch\lib\cufftw64_10.dll /y
1 archivo(s) copiado(s).

C:\Applio>pause
Press any key to continue . . .

thats what shows in the cmd when i run the patch bat

[simple ore]

looks good

[astral tangle]

so now i edit the file and then run the last bat?

[simple ore]

yes

[astral tangle]

i thought the fact that it didnt find the specefied folder was an error

Traceback (most recent call last):
File "C:\Applio\app.py", line 27, in <module>
import rvc.lib.zluda
File "C:\Applio\rvc\lib\zluda.py", line 1, in <module>
import torch
File "C:\Applio\env\Lib\site-packages\torch_init_.py", line 270, in <module>
load_dll_libraries()
File "C:\Applio\env\Lib\site-packages\torch_init.py", line 266, in _load_dll_libraries
raise err
OSError: [WinError 126] could not find specified module. Error loading "C:\Applio\env\Lib\site-packages\torch\lib\cublas64_11.dll" or one of its dependencies.

Press any key to continue . . .

that happens when running the run-applio.amd.bat

[simple ore]

show the screenshot of env variables with the path

[astral tangle]

cant send screenshots here

should i dm you?

thats the one i made

[simple ore]

I said

Path

not some random ass entry

click, add new entry

[simple ore]

No

double click "Path"

[astral tangle]

added

[hallow thistle]

Raj who?

[median monolith]

Recommendations for a place where I can ask about more technical things related to image generation?

Since this channel is more focused on rvc/ai voice stuff

[hardy yew]

is that some weird bot

what's going on

<@&1159293140440723499> there's some weird bot above that greets all incoming users

[low shard]

could you elaborate

[hallow thistle]

There's #🏙│ai-images but people there usually upload their generated images rather than actual conversation.

[low shard]

https://github.com/AIHubCentral/docs/commit/ec6cdfee0d34a0aa32db33774d0bd8de218f2b8f

@simple ore https://docs.applio.org/getting-started/installation/#amd-gpu-support-windows / https://github.com/IAHispano/Applio-Website/blob/main/apps/applio-docs/src/content/docs/getting-started/installation.mdx may need an update for the AMD GPU on Windows part, I saw it pinning to older PyTorch Versions and being lastly updated 9 months ago

[simple ore]

pytorch version is the same

[low shard]

it pinned to torch 2.3.1 while https://github.com/IAHispano/Applio/blob/main/requirements.txt uses 2.7.1, but anyways I was just saying

Did Dione Launcher die? I was just thinking about adding it to the AI Hub Docs as an alternative easy way to install Applio

[simple ore]

zluda uses 2.3.1 cu118

because I think it is the latest cu118

nvm, there's cu118, but I have not tested it

looks like 2.7.1 as well

[sharp light]

Test worked!

[low shard]

what?

[bleak ermine]

Is there an AI that can help me edit two images into 1, Like if i have a flag and there is an image with another flag as the background can i just tell the ai to change flags is all, This is an example only i need an AI name

[finite wind]

maybe the reason why one of my super high quality audio dataset model turned out to be bad model is

every audio file had a different bitrate

even though there are all 48k

when I look inside them it's all irregular

[hardy yew]

irregularity sounds bad indeed

[finite wind]

yeeah it was high enough like beyond 700kbps so I assumed it would be okay

but it was all over from 500 to 1500 ish

seems like 786kbps is the norm for wav

but it do go higher for some audios

[hardy yew]

wav is uncompressed so it's bitrate is just format-dependent and will always be the same

48000khz*16bit=768'000

[finite wind]

I swear I got this 48k wav and it's 1152kbps

[hardy yew]

24bit isntead of 16bit

[finite wind]

I gotta ditch this 24bit one instead

rest of em are 16bits :/

[analog obsidian]

dont use 24bit for rvc

[hardy yew]

no issue with downsampling 24 to 16 though, is there?

[finite wind]

WHY IS IT NOT IN THE AIHUB DOC

[analog obsidian]

dithering can mess up the model real bad, but even if the 24 bit file dont have dithering, then the noise that comes with these type of files will still have bad effects in the model

[finite wind]

I already started one with 24bit already

man

[analog obsidian]

idk... but i'd rather not use anything that is 24bit

[hardy yew]

https://tenor.com/view/spongebob-patrick-star-noted-notes-gif-17474838830648097856

(not really, although i actually should note all such things xd)

so much small info like that around in a couple servers

will definitely forget eventually

[analog obsidian]

probably a denoise should fix any 24 bit introduced issues

the problem is that 24bit comes with some very low level noise rvc learns

[finite wind]

I think I experienced that with mine actually

[analog obsidian]

best case it's going to make the model sound a bit noisy but nothing else

worst case might introduce artifacts

[finite wind]

you think it's just a simple RVC limitation or

[analog obsidian]

rvc limitation tbh architecture is very old and sensitive

[hardy yew]

almost like me

[analog obsidian]

lmao

[finite wind]

https://tenor.com/view/shh-cats-gif-10955612

[low shard]

please elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, AI Covers, E Girl Trolling / Catfishing or Roleplay
• the tutorial link used

[viral mason]

What gpu do you have (Nvidia or AMD) and what are you using it for, just curious

Die please ❤️

Thank you

[tame oracle]

New worm pfp

what a day to be alive

[viral mason]

I thought it looked pretty

[viral mason]

:(

why

<@&1159293140440723499>

why can't people that join here be normal 💔

catfishing is related to scamming, and it's in the rules that you shouldn't catfish

it's also weird lol

[low shard]

deleting it won't matter, you aren't allowed to be here anymore

[viral mason]

they already left dw

[low shard]

never here, we hate it

[low shard]

the other was there, and none can comeback

jesus

[viral mason]

?

[sly bough]

Hi, I’m looking for a voice from a random French people but I can’t find anyone who could help me please

[finite wind]

https://tenor.com/view/skeptical-futurama-fry-hmmm-i-got-my-eyes-on-you-gif-17101711

[mental stag]

Hello, I'm sorry to bother you. I admit I wasn't sure where to ask this, but here goes. I'm a Windows user with an AMD graphics card, and I have a relatively simple project, though it's not quite that simple. Basically, I want to create a Docker image, and I'd like it to use my graphics card and not my CPU. But for some unknown reason, even though I'm using https://hub.docker.com/r/ollama/ollama (the latest version released recently), when I check the performance, I see that it's using my CPU and never my GPU. I admit I'm a bit lost because I thought this version was maintained by AMD? For those curious, I'm using this Docker compose.

    build:
      context: .
      dockerfile: Dockerfile.ollama
    container_name: ollama
    ports:
      - 11434:11434
    volumes:
      - ollama_data:/root/.ollama
    devices:
      - "/dev/dxg:/dev/kfd"
    security_opt:
      - seccomp:unconfined
    environment:
      - OLLAMA_DEBUG=1```

[finite wind]

lord have mercy 48k audio with 3072kbps

[hardy yew]

I'd guess stereo 32bit?

[finite wind]

yeeep

inconsistency is real

[viral mason]

convert the audio to mono using audacity or smth

[simple ore]

Docker is Nvidia exclusive

[abstract comet]

anyone able to help me with what to do on the training section of applio collab?

ive completely forgot how to train a model properly

I have my dataset ready

just cant remember what paths I have to copy etc etc

[mental stag]

Is there no way to get past this limit ?

[simple ore]

there's a lot of tinkering required in getting AMD pass-thru working

[mental stag]

Even if I were using Linux and not Windows, it wouldn't be possible ?

[simple ore]

Windows 11 → WSL2 Ubuntu → Docker Desktop (WSL backend) → ROCm inside Linux container

if you have linux pc it is a bit easier

but you dont need docker then

just rocm pytorch will do

with linux you install amd kernel driver, pass the devices to the docker container

ubuntu 24.04 + rocm 6.4

[mental stag]

okayy thank you very much !

[brave quartz]

Any recommendation like I wanna the best one you know especially if I can add emotions to my rvc model with tts

[chilly furnace]

me me me

[low shard]

you’d need to manually use another tts and use the output in applio

[hardy yew]

A bit late to the party, but what's your issue exactly?

The process is quite straightforward and it's more or less the same in the cloud as locally

Oh, it's colab, not kaggle

So I don't know about the particular input/output path usage

[upbeat wolf]

I’m thinking about getting into AI engineering—where should I start? I’m currently doing web dev, so I have some background in JS and coding

[median monolith]

What happens if I use a model's pth with an index file of a different model when inferencing?

[finite wind]

does 96k audios even actually reach 96k in the spectrum

something tells me that 96k is a lie

SO IT WAS A LIE

[obtuse summit]

hi, i need help for an rvc model, i have an alt for trolling but the only thing is missing is the rvc, since im italian there isn't an rvc model for caine in italian, i have applio, windows 11 home, a 5060 (8gb vram) 32gb of ddr4 ram 1tb of space (2 u count the second disc) and a ryzen5 5600x

[hardy yew]

If the sample rate is 96k then 48kHz is the max frequency you can store

This is pure math

You can google Nyquist Sampling Theorem

But in short, imagine having a 96kHz sequence of samples switching between 0 and 1

That's effectively a square wave with frequency of 48kHz

Because the wavelength spans 2 samples

[obtuse summit]

i just found replay from weights and now its traning my model and it found automatically the best epoch amount :/

[finite wind]

what I'm saying was it didn't even reach beyond 24k in the spectrum in the screenshot that I posted

my bad for not specifying ofc we know how to read em

I thought I messaged here numerous times that I know that digital audio samples does not 1:1 ratio to what you see in the spectrum

[obtuse summit]

its slow so i stopped it, it takes 1m for epoch

[finite wind]

maybe that audio was originally recorded or processed at a lower rate idk

gotta seeing 96khz audio with barely reaching 24k sepctrum was not what I expected

[hardy yew]

Oh, my bad then, misunderstood the question

[hardy yew]

Hmm dunno about that really, curious

[hardy yew]

That's what I would expect TBH

But not sure

I think it's mostly beneficial in terms of further processing/filtering. Even if the mic signal remains low in the frequency spectrum, higher sampling rate will still record it with more details

But honestly, (despite the fact that I'm clearly not too knowledgeable in this specific area) I think it's for some very specialized purposes

And in general doesn't change much

[finite wind]

yeah I agree, just seeing the spectrum above 24k is an absolute void without even any interference like the below 24k in silence I just had to ask

[proven pecan]

Hey everyone.

My dad had one of his vocal cords removed due to an extremely rare cancer. Only 10 people are known to ever have had it, and he and only one other surived it 10 years out!

I'm looking to find a way to use AI to fix his voice for live speech. I tried RVC a year or so ago, but it wasn't great. Laggy, and missed words a lot.

Any newer/better products I can check out? Even if it means paying? He doesn't mind havint to buy an RTX machine if that would help.

I have probably 45 minutes of his pre-damaged voice recorded from the early 2,000's, so quality is OK.

[frail yoke]

guys
i have replay
and tried training a model
but i keep getting a cuda error
is this really because weights shut down?
because in previous versions
you'd need a weights account
pls help me out

[abstract comet]

Hey, elevenlabs did something similair for a guy who couldn’t sing anymore

https://www.youtube.com/watch?v=a12kzKiWbbQ

RVC is pretty good though, what words is it missing?

[proven pecan]

I wonder if a throat mic would work better for him

[abstract comet]

Words in a song?

[proven pecan]

His voice sounds like a loud whisper

[abstract comet]

Oh are you looking for live real-time conversion?

Seed-VC by Plachta has live conversion I believe

[proven pecan]

Yes, live voice, has to be low latency

[proven pecan]

Do you know how this works? Is is cloud based? If so ,that will likely hae too much lag, or no?

[abstract comet]

@proven pecan https://m.youtube.com/watch?v=u8mmn9lOIyw

[abstract comet]

You can download it from GitHub

https://m.youtube.com/watch?v=wLBwyG5UJ0I

[proven pecan]

Looks REALLY laggy

[abstract comet]

Ah sorry

[abstract comet]

Maybe this is good?

https://m.youtube.com/watch?v=kGbm1w9JH8A&pp=ygUVTGl2ZSB2b2ljZSBjb252ZXJzaW9u

[viral mason]

You should be using applio, it's much better

[viral mason]

What kind of trolling? That kinda stuff isn't really allowed

[viral mason]

It would sound a little bit different but would still work

[viral mason]

It depends on what you have gpu wise, if you have Nvidia you would use Vonovox for realtime, if you have AMD use Wokada tg fork

But for training use Applio, it uses rvc but it's gotten better due to pretrains and such

[viral mason]

Don't train on weights, use Applio

[viral mason]

These are old, don't suggest anything from YouTube as they're all outdated

Both you sent aren't any of the current real time options that are suggested now

[abstract comet]

Ok

Please do let him know then

[viral mason]

I'm hoping he didn't download either

[viral mason]

For future help with voice changers btw just use this

-rt

[patent trellis]

🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Local Guide

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Local Guide
• Cloud Guide(s)

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Local Guide

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

• Local Guide
• Cloud Guide(s)

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

[viral mason]

It shows the current list of the most recent and still up to date voice changers

Vonovox is Nvidia only

But is the best one

Wokada tg fork is mainly what I suggest for AMD users but also has many other options for it that I'm not entirely aware of

[abstract comet]

@proven pecan

[abstract comet]

@low shard I have trained an rvc model and am using Hyper RVC - effectively a fork of AI Cover Maker

where do i put the pth file?

i've put the index file in logs and it's been picked up

but when i put the pth files in a folder called weights in Assets folder, I can't see it

[native quiver]

Has there been any update that improves the audio/pronunciation quality of the models in recent months? I haven't used it in a while.

[abstract comet]

not that i know of

[median monolith]

Interesting, so who knows, theres potential for experimenting.
I wonder what will happen when I use the pth file of a short dataset model with the index of a really developed one, and vice versa

[hardy yew]

it's more about the differences between features in the datasets

although obviously more data usually (more or less) means more detailed feature index

[mystic ice]

What is the best video to motion capture open source repository

?

[abstract comet]

Is this the latest recommended model to use?

On the ai cover website it recommends

[nocturne mural]

https://docs.google.com/document/d/17fjNvJzj8ZGSer7c7OFe_CNfUKbAxEh_OBv94ZdRG5c/edit?tab=t.0

[hardy yew]

:OO

https://tenor.com/view/steal-racoon-cute-funny-animals-cat-food-gif-15431336

[nocturne mural]

https://media.discordapp.net/stickers/1176260690646937812.gif?size=160&name=waoowowo_%2Fxdream&lossless=true

[abstract comet]

[frail yoke]

how to i get applio?

nvm i got it

thanks tho ^^

[delicate cradle]

yea

https://discord.com/channels/1159260121998827560/1341216399372062823

there

its goated 🙏

?? im being fr

download the model

its just a way of hiding girl models

hes hiding it

the actual girl voice is in the link

yea yea

[viral mason]

please don't help those people who want to troll with girl models

it's not allowed here

if they ask for a girl model ask why

only help them if they say they're trans or smth

it's not breaking any rules for people to make them, but frowned upon

you're using them wrong

this image is a shining example of why I hate them

and why I am acting how I am towards you

most normal people that don't have brain defects are using good models not egirl ones

then what are you going to do with it, why pretend to be a girl

that isn't specific

and I find them disgusting

idk, what do you even do, how is it even fun, what's the point

I genuienly don't understand this

because I wasn't dropped 💜

being seriouse tho I want to understand the point of this

like

why?

what makes it fun

what do u do

?

[viral mason]

I don't get what you mean

is it like trolling in vrchat where you go around and say racist stuff

is that trolling

I don't understand what trolling is💔

I feel like I am talking to an alien

not in a rude way

like, I am not understanding it at all

I would never stoop so low

I understand that

but like

what do u do

how does one troll 😭

https://tenor.com/view/tole-cat-cute-gif-12080171459357821404

no gif perms :(

I do yea

I post almost all the time bc I don't have a life

if only it made me money

true, nobody is hiring me tho

getting a job now is like finding gold

just doesn't work

USA

[low shard]

I haven't really used Hyper RVC

[delicate cradle]

used to be able to

[viral mason]

I've applied to so many places and been rejected by them all 😭

[low shard]

not for e girl trolling / catfishers, those can go out here

[viral mason]

maybe I should stay a chud

[low shard]

and most of those get all out of here in fact

bye

[viral mason]

lost that when I lost my past partner

[low shard]

also that's sad 😭

yk you can ping mods when there's weird drama going on

[viral mason]

I wanted to try and understand the creature

it went as well as i thought

beep blorp boop

basically a back and forth between "what is trolling" and they say "just troll"

cannot define it or explain it

[low shard]

seriously just alert the mods, it was more like just ragebaiting

[viral mason]

most likely yea

further communication of these entities will be haulted

[low shard]

also the "im doing just e girl trolling for fun" guy:

that's why it's considered catfishing

[viral mason]

bro must be lonely

money, but no friends

[low shard]

yeah there's no need for that 😭 it was going on for a while from what i see

[viral mason]

they got two screenshots for the hall of shame

[tiny coyote]

Hey all quick question, what do you guys use to generate the sample audio clips for voice models?

[viral mason]

applio

[tiny coyote]

Ty!

[marsh tapir]

anyone know of any good decent AI image generation websites please? im so done with ChatGPT

[covert coral]

Gemini

[iron merlin]

can someone tell me where i can download the ai voice changer for my computer? I have a 50 series graphics card, I know some of them wont work for me

[crimson hinge]

how do i install llama cpp

pls

nvidia 4060

[viral mason]

I'll get u the links rq, what are u using it for btw, just curious

[iron merlin]

Just got a new computer and wanted to see if it will run good, mabye just use it with some friends

[viral mason]

arighty

https://huggingface.co/dr87/vonovox/resolve/c8034f5f6d50648a8109bb4f847182362e2b779b/Vonovox_beta_17_11.zip

https://software.muzychenko.net/freeware/vac470lite.zip

[iron merlin]

thank you 🙌

[viral mason]

you're welcome! if u need help just let me know,
exctract both then run setup64 for vac lite, then install driver
for vonovox run setup

[iron merlin]

still extracting the vonovox file xd but thank you again

[iron merlin]

hi just extracted both of them and ran setup 64 for vac lite, I opend vonovox but dont know what to click

[viral mason]

wdym?

[iron merlin]

could i DM you what i see?

I ment i opend the vonovox file, And dont know what to open

[visual citrus]

help-

[viral mason]

sure

[visual citrus]

where do i download it

[iron merlin]

nevermind, figured it out, thanks tho

[viral mason]

download what

[visual citrus]

the rvc voice changer thing

[viral mason]

what gpu do u have? (Nvidia or AMD) and what are u using it for, just curious?

[visual citrus]

amd

[visual citrus]

like does it matter

[jade flint]

hey uh how do i not hear myself w the voice changer or does it js do it like that automatically

[viral mason]

the gpu is very important to witch one u can use

[nova spindle]

Is there any new ai voice changer now that weights.gg is gone?

Im kinda looking for something like right now and easy and free

@tame oracle I see ur a helper. Can you help me?

[tame oracle]

Depends on what kind of GPU you have

also curious what you're gonna do with it

[nova spindle]

Ai enhance some 8bit vocals

[viral mason]

Weights was never able to be used as a voice changer

Are you meaning making covers or smth

[nova spindle]

I mean like taking vocals and changing it to sound like someone else

[viral mason]

Ah, easiest way to do that for free would be Applio, which can be used both on your pc locally or via a site like on Kaggle or colab

You have to remove the vocals from the music tho before converting it

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[viral mason]

This is the best way to learn how to use applio btw ^^

@nova spindle

[hushed sinew]

how to use the ai voice changer guys?

[rough lark]

is there any free alternetive to sora ai?

to create ai videos

[low shard]

elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

this is a general ai discord server

[low shard]

elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[midnight fern]

Guys is there a free motion control software like the one thats as good as on higgsfield i dont wabt to pay for shitty credits

[winged adder]

Differenze Speed using local vs colab llm model?
Localy hit 35 t/s
Using colab t4?

[low shard]

You could research the LLM speed on t4 as it depends on the model

[proven pecan]

How does it compare to respeecher?

[halcyon bridge]

I extracted the RVC-GUI straight from the github and ran the bat file, this is what i got in the cmd window

D:\RVC-GUI-Windows-pkg>python.exe rvcgui.py --pycmd python.exe
Traceback (most recent call last):
File "D:\RVC-GUI-Windows-pkg\rvcgui.py", line 4, in <module>
import soundfile as sf
ModuleNotFoundError: No module named 'soundfile'

D:\RVC-GUI-Windows-pkg>pause
Press any key to continue . . .

[simple ore]

if you downloaded just a 200kb zip with the code, yeah, that's what gonna happen

on the other hand, using RVC GUI in 2026 is wild

[halcyon bridge]

I just looked up how to make a song cover and a video on rvc popped up

Youre telling me there are better alternatives?

[simple ore]

In AI land a project from 2023 is as good as Windows 95 now.

[hallow thistle]

When the "Applio RVC" walks in.

[rough lark]

Is there an free sora Ai video thing out there?

[low shard]

RVC GUI is outdated

elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

sora AI seems to be getting discontinued soon, and there's no free unlimited ai video generated, it costs too much to run the server so they would just go bankrupt

[rough lark]

But is there a good limited one like 10 uses per day and such?

[halcyon bridge]

• integrated radeon graphics on an amd ryzen 5 PRO 4650G
• windows 10
• ai cover of a song
• https://youtu.be/0nTfLpqfN1Q?si=FJiF2RKSuKID7Olx

[low shard]

most youtube video tutorials for RVC are outdated

uninstall everything

also your PC is way to weak to run local AI

it' s better you check https://docs.aihub.gg/rvc/cloud/applio-cloud/ or https://docs.aihub.gg/rvc/cloud/aicovermaker-cloud/

[low shard]

I haven't tried much but heard about Hailuo AI and Kling AI

[rough lark]

No login and singup bs?

[low shard]

Unless you run Local AI on your own Beefy PC, you will never get a free unlimited generator without an account, especially for Videos which are the most expensive

Hailuo and Kling will require an account iirc, but I haven't tried them much

[finite wind]

holy

I stumble upon 192khz 6144kbps audio today

I had to take off my hat to pay respect to whoever have that recording devices to pull this off

[viral mason]

I've never even heard of it, probably is old tech
Vonovox is currently the best to use for Nvidia

[halcyon bridge]

i did everything according to the AiCoverMaker cloud, it gave me an error just as it was outputting (without ngrok)

EOFError: Ran out of input

at the end of the line

[halcyon bridge]

after a bit of wrangling, it worked

[agile violet]

Wondering if anyones open to doing commissions? I've tried to train ai voice models in Mango RVC but my specs and other tech issues are causing blue screens when doing so. (No longer doing commissions)

[hardy yew]

you can request in #1159289738314919936, but if it's the hardware that causes issues for you, then you might consider Kaggle/Colab etc.

Kaggle offers free 30h of GPU access per week

plenty of time for free AI training

[agile violet]

I understand

thank you

[cerulean geode]

is there like something more modern then the "Okada" voice changer?
like new ish? or what's the lastest? :D

[hardy yew]

Nvidia -> Vonovox
AMD -> Tg develop w-okada fork

-rt

[patent trellis]

🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Local Guide

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Local Guide
• Cloud Guide(s)

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Local Guide

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

• Local Guide
• Cloud Guide(s)

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

[clever garnet]

Does anyone know any github codes, hacks, or methods in general that automatically do Rosetta Stone for you?

Really apprecite some help I tried the Rosetta Stonks github chrome extension but it does not work anymore and has been patched

[low shard]

Paid RVC Voice Model Commissions aren't allowed, only free requests

also Mangio RVC is outdated, use the ai hub docs instead of youtube tutorials

[low shard]

This is a General AI Discord Server and there are many voice changers, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, AI Covers, E Girl Trolling / Catfishing or Roleplay
• the tutorial link used

[agile violet]

understood, I apologize

[finite wind]

<@&1159293140440723499>

[proven hill]

whats the issue

[wide oasis]

https://tenor.com/view/i-have-been-summoned-gif-19881033

I have been summoned

[wide oasis]

What's up

[wanton gulch]

I wonder

[wide oasis]

Well I'll be off then

https://tenor.com/view/guy-montgomery-taskmaster-nz-running-paul-williams-taskmaster-gif-4822320436128162613

[strong hinge]

hi i need help

[finite wind]

HEY I DIDN'T KNOW EVERYONE WOULD COME

there was a hacked acc and that was all

[wide oasis]

You tag the mods and the mods shall answer

[dawn copper]

Hey all. Having a weird issue with RVC Client (v2332)
I'm on an AMD card (RX 9070, 16GB VRAM) and after a while of chatting with my friends in discord they said it sounds like my AI voice profiles are "in a box". Sure enough if I listen to it in discord, the audio quality sounds very muffled, and like you're just way too close to the mic. But monitoring it in RVC, it sounds fine, and recording in audacity also sounds fine. Just seems like it's discord, and only seems to get worse the longer the chat goes on.

I use VB Virtual Audio Cable to direct my RVC over to being a mic in Windows. And what's odd is that in the RVC web client, if I change my Output away from the VB Audio Cable, to anything else.. Then back to the virtual cable? It works fine again!.. For a time. Then slowly the quality starts to degrade again, until it just sounds muffled and low quality. Model doesn't seem to matter!

[viral mason]

Helooo! could you provide a screenshot or link to where u got the version of the realtime voice changer you're currently using?

[dawn copper]

https://github.com/deiteris/voice-changer

[viral mason]

ah you're using deiteris

[dawn copper]

No idea where I first found it, but it works with AMD

[viral mason]

have you tried using Wokada TG fork?

[dawn copper]

I have not. It's kinda hard to find all these! I saw Applio was a thing, but didn't seem to support AMD when I tried it. I'll look up Wokada TG though and give it a try.

[viral mason]

one sec while I get u the link

don't want u downloading the wrong version

you use windows btw right?

[dawn copper]

Yup! Windows 11.

[viral mason]

alrighty! :3

[narrow coyote]

what ai is this?

[viral mason]

alsooo do u have vac lite or VB cable?

vac lite would say "Line 1" and vb cable would say VB smth output or input

[dawn copper]

I just have VB Cable... I think. Not sure what vac lite is!
Ah, then yeah, VB Cable.

[viral mason]

post malone, the video doesn't play for me

[dawn copper]

Oh. I think? Idk! Mine says,

[viral mason]

ye that's def vb cable

[dawn copper]

Thought so! But the "in 16 Ch" threw me off.

[viral mason]

vac lite is recomended over it due to vb cable causing weird issues on windows sometimes

https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-dml.zip

https://software.muzychenko.net/freeware/vac470lite.zip

[dawn copper]

I'll try Wokada TG, and if I still have issues I'll try Vac lite.

[viral mason]

oki doki

[narrow coyote]

oh damn, the vid needs to play for you to know

its like really realistic

i cant find it anywhere

is there a way i can send it so it does play?

[viral mason]

not sure, I'll try downloading it

[narrow coyote]

okay, if you cant tell from that one clip i have full songs and stuff with that ai that i can send

[dawn copper]

Thank you for the help!
And the Post Malone thing does work for me. Discord being weird probably.

[viral mason]

yea

I downloaded it and it worked

I can't really identify the specific model tho just by how it sounds

https://discord.com/channels/1159260121998827560/1175430844685484042

I'd look here if I was u

[narrow coyote]

The thing with this ai, it makes full songs clearly and full backing vocals with the model

its so confusing

it cant be nothing on huggingface, jammable etc

[viral mason]

and if u cannot find it there look here https://voice-models.com

[viral mason]

the music and background vocals were most likely added in as well as the slight autotune I heard

[narrow coyote]

ive been told that its not RVC, its something better

[viral mason]

oh

if it isn't rvc I have zero idea what it could be

[narrow coyote]

Can i send you some full songs that i was scammed for privately using the model?

Oh

no problem, thank u anyways

[viral mason]

copyrighted material can't be sent here sorry

[narrow coyote]

Can I ask how it would be copyrighted?

If the whole song is AI?

[viral mason]

unless it's original music or just the vocals then it's ok

[narrow coyote]

The whole song is AI generated, lyrics and everything, just post malone ai sings it

[viral mason]

ah

probably put that in the media channel

unless I had a site name or anything like that tho I wouldn't be able to tell ya

[narrow coyote]

All good, thank you anyways.

[visual frigate]

hi

I might be able to fix your mic

want to join call one?

[light quail]

Эксперемент с буффером сделай

Может поможет,

[viral mason]

this is super old, why are you using an egirl model?

wdym by this

what does that mean

but why not as like

something cool

like Goku or Darth Vader

why a woman

it's really weird to use egirl models specifically tbh

how is it fun

oh yea

<@&1159293140440723499>

egirl trolling is in the rules

it's one of the things that's really now allowed

it's related to catfishing and scamming witch is bad

[tame oracle]

but you intend to use it that way

thats the issue

[viral mason]

👀

just be normal❤️

no egirl = good healthy brain

[tame oracle]

E girl trolling is related to catfishing my guy

its not intended to be used for things like that

[viral mason]

it's not technically breaking any rules to make the models, but frowned upon

up to the mods

[cunning zodiac]

skebob

[viral mason]

https://tenor.com/view/ahoklollmao-no-way-shocked-surprised-omg-gif-16935646637611070933

[paper hinge]

hello I want to train an audio with rvc, but I can't do the second step

[viral mason]

are you using applio?

[paper hinge]

The Mangio-RVC-Fork

[viral mason]

that's quite old

[paper hinge]

is applio free?

[viral mason]

mhm! and it's able to be used Locally and on a website like google colab or kaggle

[paper hinge]

thank you I will search about it

[viral mason]

no need!, all the info is right here

-rvc

[patent trellis]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[viral mason]

anything on yt may be outdated or not very useful info

[paper hinge]

where do you advise me then to get information?

[paper hinge]

here?

[viral mason]

right there is all you'll need tbh, plus asking here ^^

[hallow thistle]

You didn't reply to a deleted message, and that might confuse some other moderators here.

Using older W-Okada version is at least acceptable, but using an E-girl voice model is a much bigger disappointment. Didn't you realize what potential consequences would be after this?

[hallow thistle]

Bidi?

[orchid vortex]

Good evening everyone, I need help because I'd like to be able to write in certain chat rooms, but I can't because the text box isn't there.

[hallow thistle]

This is literally help channel, bud. You can just explain about your issue here.

[orchid vortex]

My problem is that in some chat rooms, there isn't a text bar to type in; instead, there's this phrase: "Have you finished reading? Take a look at the # chat room." I'd like to be able to type

[pseudo elbow]

that sounds

[orchid vortex]

?

[viral mason]

Very sound yes

[white shadow]

W-okada has been doing this weird thing where it cuts off what im saying at the end of the last word and then finishes the word when I start speaking my next sentence

[viral mason]

Don't forget to @ helper role 👀

[white shadow]

<@&1159293204038955078>

[white shadow]

you right you right

[past furnace]

yo i cant hear my voice changer in game can anyone help

<@&1159293204038955078>

[chilly furnace]

me when i'm sounding

[night ravine]

when i use the voice changer it keeps on like cutting in and out like my sentences get chopped up is it a gpu issue

[low shard]

elaborate:

• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[white shadow]

My gpu is an AMD Radeon RX 6600
and I'm just trying to use Deiteris' W Okada Fork like I normally do but the past week or so its been cutting off what im saying at the end of the last word and then finishes the word when I start my next sentence.

[white shadow]

.

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

[low shard]

what’s your pc os? Are you trying to do ai covers, e girl trolling / catfishing or roleplay?

[white shadow]

wtf is pc os

[white shadow]

w-okada is a voice changer

[low shard]

PC’s Operating System

[white shadow]

windows 11

[low shard]

Yeah i understand, but there are different programs for different use cases, please elaborate what i asked

many people confuse the programs

[white shadow]

how else would I elaborate

im trying to use characters from games to voice troll in vrchat

[low shard]

are you trying to sound like goku for example or e girl trolling?

[white shadow]

no no not catfishing

nasty

things like goku yeah

[low shard]

Delete wokada deiteris frok

Try wokada tg develop

-rt

[patent trellis]

🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Local Guide

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Local Guide
• Cloud Guide(s)

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Local Guide

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

• Local Guide
• Cloud Guide(s)

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

[white shadow]

tg is full of issues

[low shard]

wokada tg develop got more recent updates, what issue are you facing? It’s way more suggested

[white shadow]

well the past 5 times ive tested it, it gives errors when I try pressing start, images dont work when trying to add one to a model, quality isn't as good as the Deiteris one

plus another bonus. Its not an error but the ui just doesnt look good to me

[low shard]

Could you please try it again and show me screenshots of the error and the settings? I could help you out

Wokada deiteris fork isn't going to be maintained and suggested anymore

It's kind of treated like Mainline/Original RVC

[white shadow]

alr hold on lemme try again

cant send the error cause its too long

you want it in parts or will it look to messy

[white shadow]

in the mean time heres the settings

welp. Just tried restarting it and now it wont open

holy issuessss

[low shard]

why did you choose server instead of client?

it's not needed if you aren't going to use it with wasapi or flexasio, which you aren't doing

The difference between Server and Client Audio in Original Wokada, Wokada Deiteris Fork & Wokada Tg-Develop Fork:

• Client: easy to use, can use echo, sup1, sup2 cancellation
• Server: harder to use, can significally reduce delay if used with Wasapi or Asio

Also VAC Lite is more suggested than VB Audio Cable for Windows

[low shard]

show the cmd logs, the UI doesn't show much

[low shard]

many people use wokada tg develop instead of deiteris, i feel like it's just that you aren't used to the new UI, which btw you can swap back to but it could miss some new features (like the voice effects)

[rough lark]

What would you guys consider to be the best free Ai for pictures

Btw @low shard I tried the 2 ai's u recommended for video gen but both of them need credits what I need to pay for

[fierce bone]

stupid question: To your knowledge, has there ever been any successful use case of RVC with AMD cards? Like the RX 9xxx series? I've been out of the loop for a while.

[fierce bone]

Not directly, no. But you're welcome to reply.

[hardy yew]

TG develop w-okada fork has a DirectML version that supports AMD GPUs

it seems that it does work for people usually

although i've seen some people having issues, too

personally haven't ever had a chance to try, Nvidia user here

[fierce bone]

I've a 2060s that needs repair for the moment, but current build uses rx 9070xt for gaming, although I can easily set up environment for testing it out. So anything is welcomed. I noticed AI Hub Docs mentioning Applio Zluda which do seem interesting.

Have to take a look at TG develop w-okada fork too it seems

[hardy yew]

https://docs.aihub.gg/realtime-voice-changer/local/tg-develops-w-okada-fork/#download-amd-intel-gpus-and-x86_64-cpus-on-windows

You can find lots of useful info here, make sure to download the (...)-dml.zip file, not cpu/cuda

This is for realtime BTW

Didn't ask before if that's your goal or offline inference/training

[fierce bone]

Both, eventually.

[hardy yew]

For offline inference/training then yeah, Applio can supposedly be run with AMD cards but it's more work

https://docs.aihub.gg/rvc/local/applio/#amd-on-windows-precompiled-fix
Perhaps this sheds some light

[fierce bone]

I train models when I have to, but atm its not possible for me until i figured out if the amd card can do at least something. x)

[hardy yew]

Can also consider training in the cloud. Kaggle offers 30h of GPU training time per week for free, which is quite nice

[fierce bone]

Yea I did consider that option.

I don't mind putting work in anywho. I don't expect something huge coming out of it eitherway haha

[hardy yew]

for me cloud is more convenient anyway, I don't have to keep my PC running etc.

but each to their own, obviously

there's lots of reason to prefer one over another

[fierce bone]

That's totally fair.

Thanks for shedding some light on it. Appreciate the kindness. 🙂

[hardy yew]

No problem. Good luck!

[fierce bone]

Thanks lad. x)

[buoyant sphinx]

Hi i have vonovox and virtuakl cable installed and a model installed but when i try run it is just says warmign up whats happening?

im on a rtx 5060

[viral mason]

have you tried closing it and reopening it?

[simple ore]

@wraith fractal if you need help you can ask here

[buoyant sphinx]

Yes and rebooting my pc

[viral mason]

strange

usually that bug is just a one time thing where once you close it and reopen it won't do it again

are you using the vonovox Beta or 1.6.9

[buoyant sphinx]

Uh

lemme check

Im using

beta 17_11

[low shard]

well not everything is free unlimited

[low shard]

this is a general ai discord server, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing or roleplay?
• the tutorial link used

RVC means Retrieval-based-Voice-Conversion, but is often confused with realtime voice changer

[low shard]

it's normal for it to take a few seconds to warm up

[buoyant sphinx]

I had it runnnig for 10 miniutes doing that

[low shard]

what's your PC OS? What VAC are you using?

[buoyant sphinx]

470lite

whats pc os?

[fierce bone]

I'll remember in the future. I got my answers covered as expected this time so no worries. 🙂 Thx for letting me know.

[fierce bone]

He's just asking if you're on Windows/Linux/MacOS or something else I think. 🙂

[buoyant sphinx]

oh windows

[low shard]

Operating System

Are you sure you updated your windows version and Nvidia Drivers via the Nvidia App?

[low shard]

ohh my bad thought you were confused between RVC for AI Covers and realtime voice changing lol, that happens a lot

[buoyant sphinx]

I got this pc 2 months ago i havn't updated the dribvervs

[low shard]

check out: https://www.nvidia.com/en-eu/software/nvidia-app/
you don't need to login to update the drivers

btw to be sure you're using the right software, are you trying to do AI Covers, TTS, E Girl Trolling / Catfishing or Rolaplay? The RVC acronym gets frequently confused

[buoyant sphinx]

Roleplay

[fierce bone]

Not at all hehe. Using RVC for training/conversion of different sources etc. Usually local and with great success for small tasks only. But current build has a RX 9070XT which weren't with AI in mind anyway. So just trying to see what the possibilities are today. x)

[low shard]

it's best you update your nvidia drivers with the nvidia app, and check for any windows update in the settings

[buoyant sphinx]

ok but i have autisim

[viral mason]

how does this contribute to what Nick said 💔

I am too but I don't see how that helped lol

[ebon tapir]

does soldier boy lag?/

[narrow coyote]

can someone help with the forum i posted please?

[low shard]

what?

[ebon tapir]

widk ssometimes when i use somew voice models it souns liek a robot

[low shard]

This is a General AI Discord Server and there are many voice changers, elaborate:

• your pc gpu
• your pc os
• what are you trying to do: TTS, AI Covers, E Girl Trolling / Catfishing or Roleplay
• the tutorial link used

[ebon tapir]

iddk allat but i wanna be soldier boy

he is tuff

whats a good gpu to run soldeir boy

[low shard]

there are different thousands of AI program, I can't know whatever are you doing

this server does much more than one single program that works for everyone in 1 click like chatgpt

[ebon tapir]

oh well

[low shard]

please elaborate everything I asked

[ebon tapir]

sorry

[low shard]

if you don't elaborate how am i used to know your issue and how to help

[low shard]

it's fine dw

[ebon tapir]

i waas js wondering if doldeir boy lag cuz butcher sounded like ehhhhhhrooo

[low shard]

i would need to know more technical details that i asked, like your pc gpu, os and the tutorial link used

[ebon tapir]

sec

my gpu is 4060

doess it work??

[ebon tapir]

can u send me a good voice changerl ink

[viral mason]

what?

[viral mason]

use Vonovox, I'll get u the downloads

https://huggingface.co/dr87/vonovox/resolve/c8034f5f6d50648a8109bb4f847182362e2b779b/Vonovox_beta_17_11.zip

https://software.muzychenko.net/freeware/vac470lite.zip

[ebon tapir]

do it got viruss

[viral mason]

no 💔

[ebon tapir]

[viral mason]

if you need any help lemme know

[ebon tapir]

im finna be soldier boyyyyyy

thanks btw

[viral mason]

no problem!

[fiery topaz]

why does my ai okada voice changer not open

[viral mason]

what gpu do u have (Nvida or AMD), and did you use a tutorial on youtube?

[fiery topaz]

rtx 5090 and no

[viral mason]

ok do you have a link to what one you downloaded?

I wanna see if what you're using is outdated or not

[fiery topaz]

yea hold up

https://github.com/w-okada/voice-changer/blob/v.2/docs_i18n/README_en.md

[ebon tapir]

sorry thowhat i ddo when i installed both??

[fiery topaz]

i did get my one 10 months ago