✨│ai-help

[low shard]

to use fuser -k, this will automatically kill the open port processes when user clicks cancel run code in the start cell, to not cause any issues if the start cell gets re-run

fixed the typo rn https://github.com/Nick088Official/Applio/commit/701995c8bc9a89c2abb38ddccf84ad78515135e0

[low shard]

https://tenor.com/view/pull-request-code-pr-github-fulp-gif-21584424

I might make an awesome tunneling list for jupyter notebooks, who knows what can happen in the future

[hallow thistle]

It looks like one SQL file went missing from Weights' database, which somehow also makes model searching unusable. The "My Creations" part still working fine, but without the creation count number.

[viscid moss]

https://tenor.com/view/bug-juice-loovoo-the-eel-arcane-gif-27671641

[coarse fjord]

is v.1.5.3.18a the latest one?

[low shard]

this is a general ai server, I’m guessing you’re talking about original wokada from an outdated youtube video tutorial, they are all old, that’s over 2 year old

[subtle merlin]

anyone can teach me how to download voice changer?

[coarse fjord]

oh sorry i didnt knew it

[subtle merlin]

so that method is still working? right

[chilly furnace]

antasma returns to ai hub

[low shard]

it’s severly outdated, don’t use it

[low shard]

it’s fine dw

[dusky geyser]

Hey, trying to export a video through the storybook on openai,

it's not using the correct videos even after clicking each one and replacing them, exporting just gives the original video before any changes.

Anyone able to explain what I might be missing

[median hearth]

Best girl rvc model

[cold stirrup]

where do I download the ai hub

[low shard]

AI Hub is a discord server, not a program

[viral mason]

have you read the rules?

[oak edge]

are there any alternatives for kaggle? @low shard

specifically can I change the applio url to a public url like the one in colab?

nvm did it myself

[low shard]

lightning.ai and google colab

[oak edge]

I transplanted google colab's gradio to kaggle

[low shard]

Applio Kaggle recently updated because of an ngrok issue, check https://docs.aihub.gg/rvc/cloud/applio-kaggle/

[oak edge]

💀 honestly why kaggle applio doesn't have gradio as an option

[oak edge]

Ooooo

[low shard]

the latest Applio Kaggle uses Gradio + LocalTunnel, that's because it needs a 2nd tunnel to show the filebrowser (needed bc kaggle built in filebrowser isn't as good as google colab one) and tensorboard (which displays automatically in google colab but not kaggle)

If you modify it to work ONLY with gradio, you won't see the filebrowser nor the tensorboard

[oak edge]

Trueeer

[oak edge]

Also what's the file browser you are mentioning about here? As in like the directory? Or am I missing some important point?

[low shard]

btw lightning.ai is also a good option, it allows web uis on free tiers without needing encryption (so no ban risk), gives around 80 hours of T4 GPU freely per month, and can use only gradio since it displays the tensorboard correctly

[low shard]

You know that 3rd public url you see that basically opens as a file explorer? The program's name to do that is Filebrowser

[oak edge]

Wait how fast is the training compared to kaggle?

[oak edge]

Never have I ever clicked on the third url, what does it even do? Why is it there?

🥲 I'm someone who rarely uses tensor board cuz I simply don't hear much difference between 100th epoch and 280th epoch (10th epoch to 100th easily recognizable though)

[low shard]

that depends on the cloud gpu being used,
you can technically use way more powerful gpus on lightning.ai for free at the cost of lower free gpu time monthly, might be better you check https://lightning.ai/pricing for more info (they give like 15 credits for free per month)

[oak edge]

I'll try to sign up in the morning I have to freshly train a new batch anyways, also can I import the dataset from kaggle to my drive?

[low shard]

like if you compare a L40 to the T4x2 on kaggle, the L40 would be faster but you'd get less gpu time

if you compare lightning.ai's T4 to Kaggle's T4x2, kaggle would be better

kaggle does offer more gpu time since it's 30 hours weekly, tho it's more risky technically

I mean, it's your choice

[low shard]

Nope, it isn't like google colab

I mean you could like download your dataset and just upload it to your drive

[oak edge]

I'll try to dig in main thing I'm worried about is data set 🥲 I wish there was some way to import my data set from kaggle to lightning directly

[oak edge]

The data would cost 4 GB 🥲 I only get 10GB per day

[low shard]

don't you have your dataset locally saved too?

[oak edge]

🥲 I did most stuff in cloud like I directly imported to kaggle from Google drive public url

[low shard]

wait so is your dataset saved in your drive currently or not?

[oak edge]

I have to dig into the lightning's ui maybe I can find a way

[oak edge]

Yesss

That's why I'm asking is there any way to directly import 🥲 instead of downloading and re uploading

Can I make my dataset public in kaggle and use that public link to directly import?

Or does direct import work in lightning from drive?

[low shard]

I think the best way would be that you modify the code to use gdown to download your dataset from your google drive
I can't seem to find any info about directly importing kaggle or google drive datasets directly into lightning.ai

[oak edge]

Ooooh they have 10 hours of a100? @low shard or can I go for h100 or l40s

[oak edge]

Maybe I'll try to dig in a bit on how the lightning's ui feels first 🥲 maybe after seeing the workspace I might get an idea

[low shard]

https://lightning.ai/pricing

I mean you don't need very expensive and fast gpus like a h200 for RVC btw 😭

[oak edge]

Okay so how fast the training speed would be in a100 of h100 of l40s,
For context my data set took around 12 Hours to reach 200 epochs in T4x2

[oak edge]

I mean this one particular project is like one time thing so I'm.not gonna probably use lighting afterwards

Or at least until a new version of branch drops

🥲 I tested spin v2 , it doesn't feel much of an improvement compared to contentvec too at least for my Tamil songs

[low shard]

are you sure you actually want to train on like h100? first time I see someone wanting to do that since they usually cared more about getting more free gpu time monthly

[oak edge]

😭 another thing is I'm going to extract and train with crepe tomorrow , bcz I feel like rmvpe doesn't get the pronunciation and overall tone correct

[oak edge]

Nah I don't care much 🥲 as I don't train multiple sets, I just want to finish my thing as soon as possible (you already know I train for like 30-40 hours per week in colab right 💀 that was really annoying so faster the better)

[low shard]

how long is ur dataset btw 😭

[oak edge]

This current one is 58 minutes

🥲

I went through audacity for noise gate and truncate as well (used applio docs not aihub)

[low shard]

I'm not really sure about the exact training speed gain from an h100 compared to t4x2, but theoretically it should be faster, the h100 has way more tflops and cuda cores

[oak edge]

Eventhough the results with rmvpe are satisfying, I feel like the pitch could be more accurate with crepe? (And better pronunciation? correct me if I'm wrong currently crepe extraction has the best quality when there is no noise in the training data set right?)

[oak edge]

Lemme google it how much percentage

[low shard]

you love to experiment i see?

[oak edge]

Yeaa 🥲 trial and error always

Idk much peeps doing this so imma do it

[low shard]

you already know I train for like 30-40 hours per week in colab right 💀 that was really annoying so faster the better
dedication fr

[oak edge]

😭😭 and honestly sometimes I wonder what can I even get out of this because that 3 hour voice data there, is just my own plain voice 💀

[low shard]

theoretically, but tbh most just stick with rmvpe: #🔊│ai-development message, #📑│making-models message

[oak edge]

So h100 seems 3x faster than T4

[oak edge]

I'm getting no access for it

[oak edge]

💀 I trained 285 epoch with rmvpe, so why not venture out into crepe for same model and see the difference (plus this time going to go with default contentvec instead of spin v2)

[low shard]

tbh I always trained rmvpe like most people

be aware to not expect too perfect results for things like non speech sounds, rvc models usually don't super realistically scream for example

[low shard]

I mean you can experiment however you want ofc

experimenting is good

just telling ya that usually rmvpe is more suggested

[oak edge]

And the male singer I'm training for have real high pitch he screams a lot and sometimes the slicing size of 5 seconds is not enough

[oak edge]

🥲 mmm true that

I will try out my best and it doesn't seem h100 being 5 hours is not that useful as it's only 3.5x faster than the t4

Which means only 1.7x faster than t4x2 theoretical, and yea it's not gonna cut the 6 hours mark even to train 200 epochs (plus also considering I would be going crepe which is significantly slower than rmvpe ? - idk heard some rumors like that, have to go with something like L4)

Welp gotta get some sleep now already 2 am here and it's freaking 36°C

[low shard]

Yeah people don't usually want to train on such high end gpus just for RVC 😭

[analog obsidian]

f0 extractors dont change the quality of the model, but they can affect how stable the voice is

[oak edge]

😭oo

[oak edge]

Stable as in more pitch accurate?

[analog obsidian]

yep

some bad f0s can make the voice sound uhh wobbly idk how to explain it, english is not my first language lol

[oak edge]

And more tone accurate ? Also feeling more closer to his real.voice assuming data set is really clean

[analog obsidian]

i have tried all of them in rvc

rmvpe is just superior

[oak edge]

😭😭 it's okay I get it

[oak edge]

But it kinda sucks with pronoucniatio. When index goes up

[low shard]

did u try dio?

https://tenor.com/view/za-warudo-the-world-dio-brando-stand-joseph-joestar-gif-26463002

[oak edge]

Swift?

Ohh YEA did you try out swift? @analog obsidian

[analog obsidian]

haha but unironically yeah, its better than harvest and pm lol

[oak edge]

Aren't these really old names?

[analog obsidian]

nah that just contentvec being bad

[oak edge]

💀 this is giving me mangio rvc memories

[oak edge]

I used spin v2

[analog obsidian]

yup lol

[low shard]

yeah, dio is a pretty old f0 that in fact got removed from applio as it's not suggested at all

[oak edge]

🥲 like both custom pretrain and extractor

[low shard]

tbh I haven't really played around with it much, but I heard some mixed, a bit more negative, opinions about it in the Vonovox server

It should be technically a bit better than rmvpe, but I can't say that it is in practice since i saw some negativity in it

[oak edge]

😭 i was too amateur back then idk how was it tbh

[analog obsidian]

ah no idea, when i tried spin v1 index it was actually way better than contentvec

[oak edge]

Maybe because it was English data sets?

For me spin v1 sucked hella bad that contentvec

Spin v2 kinda feels similar to.contentvec

But I can't pinpoint which is better

[low shard]

you surely love playing experimenting to even try pm lol

[oak edge]

As both have pronunciation issues, maybe I will bet for spin v2 at least in the lady 2 audios I tried and it kinda felt like pronunciation was good

[oak edge]

💀 what's in there I can't open it it's 2 am and I'm in bed

[analog obsidian]

spin v1 is less fuller than cvec due to some issues, but the pronunciation is still way better than cvec

i know coz i compared them too

[oak edge]

😭😭 wait how much would you rate compared to spinv2?

[analog obsidian]

spin v2 was meant to fix the fullness problem

[analog obsidian]

can't give it an opinion because no good pretrains for it

[analog obsidian]

i was bored 😭

[oak edge]

💀 ahhh maybe then Mine was undertrained when spin v1 was used?

[low shard]

is there gonna be like a legacy core pretrain for spin-v2 btw?

[analog obsidian]

noobies v2 spin pretrain has degradation,i showed him

[oak edge]

What about the current one I think @simple ore mentioned

[low shard]

[analog obsidian]

this is degradation

[oak edge]

💀💀💀ಠ⁠_⁠ಠ

[oak edge]

Holy shit gng again I can only hear this tomorrow I will take a listen

Degradation means the voice getting more metallic away from original data set voice?

[analog obsidian]

the voice starts to sound very raspy and broken

like a demon

[oak edge]

Ex-fucking-actly I was feeling it around 280epochs

I had to turn on autotune

To somewhat balance it

😭😭

[analog obsidian]

im trying to find a temporal fix for the degradation problem of rvc

[oak edge]

It felt like some robotic a bit? Idk how to explain kt

[analog obsidian]

after that yea i can try porting it to v2

[low shard]

goodluck

[oak edge]

🔥🔥

So would you suggest I train with contentvec then? @analog obsidian

[analog obsidian]

yea keep using cvec and the og pretrain

[oak edge]

Thanks for it 🙏 imma try with crepe tomorrow and update how it goes

[analog obsidian]

no dont use crepe, its broken in rvc

due to the way the pipeline works or something idk why honestly, i just know it doesnt work like it should be

rmvpe was made with rvc in mind so it works as expected

[low shard]

-# it honestly feels like RVC is held by ducktape

[analog obsidian]

FR 😭

[low shard]

I saw you talking about ddsp-svc or smt before, did you also try seed-vc?

I was thinking of giving seed vc a try for fun, it's 0shot anyways so it's easier to use

[analog obsidian]

i havent tried seed-vc, but honestly ddsp-svc seems to be a great rvc replacement, same quality, realtime support, being able to train with 8gb vram gpus, fast training

it doesn't degrade like rvc

[low shard]

which branch are you talking about tho? the main one is kinda inactive since 10 months

[analog obsidian]

https://github.com/yxlllc/DDSP-SVC/tree/6.3

yxlllc is who made the realtime code of rvc iirc

and trained rmvpe

[low shard]

thx

[low shard]

i mean, hopefully one day this can 100% replace RVC

[analog obsidian]

[low shard]

I think that it's more probable we get a replacement than "v3"

[analog obsidian]

it needs a pretrain so we can train using smaller datasets, currently it's training from scratch so it needs an 1 hour dataset minimum

[low shard]

the only issue is we have already thousands of RVC models, so we are kinda stuck on RVC which isn't very sustainable (the average user mostly cares about inference rather than training)

[analog obsidian]

yeah changing from rvc to ddsp-svc is not an easy task

[low shard]

boohooh

[analog obsidian]

but it's a good alternative, members of the rvc team are involved

[analog obsidian]

ddsp-svc realtime asks for slightly more vram than rvc

but lower than so-vits-svc realtime

[viral mason]

What even is ddsp, what's the difference between it and rvc

Does it produce better models?

[minor nebula]

Error
unhandledrejection
no error stack
NotSupportedError: AudioContext.createMediaStreamSource: Connecting AudioNodes from AudioContexts with different sample-rate is currently not supported.
98/t/Un/this.setup@127.0.0.1:18888/index.js

im getting this when i put my main mic input n

[analog obsidian]

https://github.com/yxlllc/DDSP-SVC/tree/6.3

[limber hawk]

Как настроить голос из

этой группы

[sand bison]

Hi, I have a question. What is the latest Google Collab notebook available so far for training AI voice models? In my case, the last one I used to train a model was Applio in the Cloud, but it took much longer than doing it with another Collab. And I don't know if there is a newer Collab, since my PC can't train locally.

[viral mason]

I would recommend using kaggle or lightning.ai as both give way more hours for free users compared to colab

[sand bison]

or ok but lightning.ai do I need to enter any more information or just create the account with email?

[viral mason]

lightning and kaggle both require an email and phone number verification

kaggle is easier to learn tho

[sand bison]

or ok thanks a lot I'll see how I do it 😄

[somber oriole]

can anyone else not hear them self
it doesnt say my mic is moving or making noise i uinstalled reinstalled used different voices
and nothing

ive tried everything and this isnt working for no voice

its literally just making my pc lag like crazy

[tawny radish]

which w-okada version are u using

im not saying it depends on the version, im checking if ur using the correct one

[somber oriole]

the latest one let me see rq

not sure tbh i just know it was like 218

2.1.8

or sum

it has 18 at the wnd

end

u gon help me or wha

[viral mason]

what link is it?

[somber oriole]

want me to send link in priv

done

so can anb help me

[viral mason]

yea pretty sure that's outdated as hell, what gpu do u have?

[somber oriole]

i got a nvidia 5060

[somber oriole]

wait really the video was 3 momths old i alr used 2 different links

i think irs a problem else where just not aure where

[viral mason]

yea u should use vonovox as it's the best most up to date one and the only one currently still actively getting updates

[somber oriole]

can you send me the link to it

[torpid granite]

Can someone help me out Applio on lighting ai is sleeping

[viral mason]

all youtube tutorials use outdated stuff because they don't know about the new stuff and or are just being stupid

srry about taking so long I'm on vrchat

[torpid granite]

I asked Nick but I think he's busy or some

[viral mason]

use the guide for vonovox

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options 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. GUIDE

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

⚔️ Wokada Deiteris Fork vs Vonovox

For Windows Nvidia, Both Wokada Deiteris fork and Vonovox have similar performance & quality. Users should read the pros and cons for both and choose based on their differences, such as UI and Vonovox's paid effects.
Read Wokada Deiteris Fork Pros&Cons & Vonovox Pros&Cons

⛔ 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

[half umbra]

I could use a little help since my brain isn't working I guess.

Under the Wokada fork, I had it set up with Voicemeeter and Minihost. VM for routing, Minihost for noise cancellation (which works wonders).

For Vonovox, under the Audio devices, I'm missing all the other input/outputs from ASIO and the WASAPI setup is foreign to me using Minihost. Just broke on my when I tried to change it.

If I change Vono to WASAPI, my devices are there but obv Minihost doesn't work b/c it's set to ASIO.

Anyone got any ideas on how to fix/set this up without completely chanign my setup?

[somber oriole]

its no problem i was eating anyway

[somber oriole]

ye i can see that

could you send me the link to the right one tho

[viral mason]

it should be in the guide but sure!

[somber oriole]

ok thanks because i lit cant find it in the guide at all

[somber oriole]

i got vonovox like u said im so confused how this works

its different from the other and i still cant tell if it working or not

yo

[craggy bough]

[somber oriole]

yo it aint working tho ive tried 3 different apps

and none of them work can somebody please help

where is the support team in here bru 😭

they honestly mad useless

[craggy bough]

what are you using

[somber oriole]

vonovox

someone said to use vonovox cause its better

and i cant get it to work on here either

[craggy bough]

https://docs.aihub.gg/realtime-voice-changer/local/vonovox/#download--installation

[somber oriole]

yea

from there

freom the guide spot

and i pressed trhe nvidia vonovox

[craggy bough]

uh huh

[somber oriole]

it says im running latest version too

yo wsp

im here

[craggy bough]

if hes dming you its probbaly a scam

[somber oriole]

oh rly?

[craggy bough]

i havent used any of the voice changers so id just wait for nick to come online

[somber oriole]

nah no dm js a friend request

wait

r

u

using nvidia

any of u

[craggy bough]

im on a mac

[somber oriole]

dammmmmmmmmmm

is there a official tutorial

cs the guide is just full of links

where

can u send me the link

or if iuts in ther server

send it in priv then

[craggy bough]

yes you can

[somber oriole]

ye i was js gonna say u can

u js did sus

[craggy bough]

yeah the guide is just links

say goodbye scammer

[somber oriole]

lollll

u were right

is there anb i can ask

that ain a scammer bot

wait am i supposed to have vb cabled downloaded for me to hear my own voice and for my voice to work?

[craggy bough]

@low shard but hes sleeping rn

[somber oriole]

am i supposed to have vb cabled downloaded

[craggy bough]

if only you read the guide

[somber oriole]

nooo i dotn wanna use it on dc or games i just wanna heart it if it works and it didnt work

ima see rn

[brittle wing]

....

can the world just end already

[elfin nebula]

So fish speech local only works on Linux right?

Looking at the link on their github page. But I just wanna make sure
https://speech.fish.audio/install/

[somber oriole]

can anyone pls tell me what im doing wrong

i cant hear my self and the volume doesnt work

[knotty moth]

there should be docs for installation on other platforms

btw you can try another cloning TTS: #📰│dev-updates message

[somber oriole]

this is the volume setting

why isnt it working

sb pls helpp

[elfin nebula]

The links on the ai docs file for fish speech leads to their github, and the installation link on their github leads to the link that I just sent.

Alternatively F5 TTS has an installation guide but I'm not sure if it's compatible with Windows. Is chatterbox just better?

[knotty moth]

why do you go back to that old wokada version after having vonovox not working on you

[knotty moth]

chatterbox is claimed to be comparable to 11labs in some ways

[somber oriole]

it didnt work either the sound was weird and idk how to fix it since i only know how to fiz it in w okada

ans thi isnt the old version it was the most recent one

so can anb help me bru

like what settings could i have done wrong

bro pls where is the customer support at

or wtv its called

[knotty moth]

please check the aihub docs for the fork version

[somber oriole]

is that in guides

and is it w okada

[craggy bough]

already sent it but he doesnt wanna read

[somber oriole]

u did?

sorry bru i was prolly fixing chrome up

i didnt see it

ok ima uinstaill this one

where do i go lol

i pressed the link in the guide

mvm found it

downloading rn

said it was the latest version

[half umbra]

Any possible assistance with this?

[elfin nebula]

Interesting... I've attempted to install chatterbox and this error showed up. Is there anything I should do? Also, if I get passed through the installation would this open up a local tab?

[simple ore]

try cloning the repo and installing that

[heavy cloak]

when i try to use a voice changer the voice sounds so robotic and cuts out how do u fix it?

[simple ore]

python 3.11 only

[elfin nebula]

I'm going to level with you, idk what that means

[simple ore]

git clone

or download https://github.com/resemble-ai/chatterbox/archive/refs/heads/master.zip and unzip

then pip install -e . from that folder

[simple ore]

dot at the end

[elfin nebula]

Sorry mb missed it

[simple ore]

hold on, there' some other issue stll

[simple ore]

yes, that

hold on

do you need Russian or Chinese?

[elfin nebula]

I just need en tbh

[simple ore]

or the multi-language model

okay, then simply pip install chatterbox-tts==0.1.2

[somber oriole]

still isnt working sb pls help and why does this error keep coming up when im pressing start it doesnt go away

[simple ore]

you'll get the original en model

[elfin nebula]

Still on the folder?

[simple ore]

delete the content of the folder

I do recommend using a virtual environment though

[coral nexus]

I got a question, im new to this vocie changer client demo. A friend of mine helped me install it, everything is set up, I have the virtual cable though none of the voices seem to be working? Is there a reasoning or something im missing? Do I need to reinstall the application? Or is there something on the appilcation I need to adjust.

[somber oriole]

anb can help rq bru

[somber oriole]

got the one u tood me abt and still dont work

[elfin nebula]

I've uninstalled and ran the command, but the folder is still empty

[simple ore]

if you did not create & activate the virtual environment, then it got installed under C:\users\user\appdata\local\python ...

[somber oriole]

dude seriously where r the mods at what r they doing

what r they even mod for if they aint helping this is wild

[simple ore]

what does the error say?

[elfin nebula]

Wait at some point you said Python 3.11 right? I got 3.10 but also, no python folder under appdata/local

[simple ore]

3.10 would probably work

[somber oriole]

that

[somber oriole]

these r my settings

and in the command setting it says this

[simple ore]

can you're type what it says, I can't read your screenshot

[somber oriole]

oh ok lol igu

an error occurrd during voice conversion check command line window for more details

and in command window it says

RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
one For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with "TORCH_USE_CUDA_DSA' to enable device-side assertions.

[simple ore]

what is your gpu?

[somber oriole]

rtx 5060

8gb

ram

[simple ore]

i suggest taking a read

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options 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. GUIDE

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

⚔️ Wokada Deiteris Fork vs Vonovox

For Windows Nvidia, Both Wokada Deiteris fork and Vonovox have similar performance & quality. Users should read the pros and cons for both and choose based on their differences, such as UI and Vonovox's paid effects.
Read Wokada Deiteris Fork Pros&Cons & Vonovox Pros&Cons

⛔ 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

[simple ore]

and installing a version compatible with your GPU

as well as a proper virtual audio cable

[somber oriole]

it was compatible the one i installed

[somber oriole]

i alr have a virtual audio cable thing i downloaded it

[simple ore]

do you not see the difference?

[somber oriole]

WHAT

can you send me the right link

or where do i find that

[simple ore]

click the 3rd link above

[somber oriole]

wokada deiteris fork?

[elfin nebula]

Okay, I found the file do I just run the installer?

[simple ore]

no

it is already installed if you ran pip install chatterbox-tts

[somber oriole]

ive never downloaded from this what do i press

[simple ore]

you download 3 *.z* files

[somber oriole]

oh so the first 3

all of them

[simple ore]

and you unzip them using 7-zip or winrar or winzip

[somber oriole]

so the firsr 3 right?

[simple ore]

yes

[somber oriole]

this gonba be a long process? since i need to download 3

[simple ore]

read the guide

[somber oriole]

wtv im downloading it rn let me see guide aswell

[elfin nebula]

Is there a tab when I run chatterbox?

[somber oriole]

oh yes i figured it out ill lyk if it works noobie thanks a lot

u should be a mod

better then the mods they have rn

[simple ore]

you can run this

change AUDIO_PROMPT_PATH=r"voices\david.wav" to the reference voice you want

[elfin nebula]

How about if I have a voice clone I wanna use?

[simple ore]

or this

as I said, change the AUDIO_PROMPT_PATH to the .wav file with the reference voice

and tts3.txt to the text file you want it to read

[elfin nebula]

Ah I see I'll try that out now, thank you

[simple ore]

you need 10+ seconds of audio

not more than a minute

it only takes 1st 10s

[elfin nebula]

I can use flac. files right?

My data set is in flac

[oak edge]

@low shard as it turns out lightning is not available in my region

[somber oriole]

it keeps cancelling my download @simple ore and says site wasnt available

is that a me problem os it like that for everyone else right now

[oak edge]

can you explain what's ddsp-svc ?

[analog obsidian]

https://github.com/yxlllc/DDSP-SVC/tree/6.3
there is a readme here

[oak edge]

sorry i just went and read

[analog obsidian]

[oak edge]

now after reading that I've got a doubt

is rvc the bestest voice conversion tool out there?

(i'm asking in terms of which has more quality)

or are there any other top end conversion tool

[analog obsidian]

ive heard so-vits-svc and ddsp-svc can beat rvc if you have big datasets

[oak edge]

that's far better than rvc

[analog obsidian]

diffsinger is also good if u want to create ur own vocaloid lols

[oak edge]

are those 2 only? but aren't both of those open source?

wait so current market's top conversion tools are all open sources?

[oak edge]

i mean i need to search up what's a vocaloid

[analog obsidian]

you mean like voice conversion as tts or in general

for singing so-vits, rvc, ddsp-svc are good

tts idk i dont use that lol

[oak edge]

i mainly use rvc for singing conversion

but not real time though @analog obsidian

i rarely use real time stuff, so I just need top notch quality for singing conversion

[analog obsidian]

so-vits and rvc are technically the same

ddsp-svc is diffusion based, so better than gan (rvc/sovits)

but uses mostly the same shit as rvc

cvec and nsf hifigan

[oak edge]

[oak edge]

also i thought about why aren't there an option to change hifi gan

there are new models at least according to applio or aihub docs?

[analog obsidian]

because the only way to fight degradation in rvc pretrains is to either getting a huge dataset or completely changing the architecture

[oak edge]

@analog obsidian

[oak edge]

as in changing the vocoder?

[analog obsidian]

so changing the vocoder requires a new pretrain

wow crazy

[oak edge]

I want to try out refine gan but i have no single clue if i can even change them

[oak edge]

are there pretrains for refinegan?

[analog obsidian]

the refinegan implementation in applio is not the refinegan of the paper, is noobies version of it

so is not "refinegan"

but noobies refinegan

[oak edge]

how do I use it then?

[hot lagoon]

is Gabox's voc_fv4 stile the best for vocals?

[analog obsidian]

ask him to do a pretrain for it (wont be possible since no one knows how to fix rvc degradation)

[oak edge]

ahh wait can you send the degradation you sent at night?

i wanna have a listen

[analog obsidian]

i like it more than fv5

yea

[oak edge]

even with pretrain I want to know how do you even change from hifi gan to something else in ui?

[oak edge]

I didn't experience this sorta behaviour this just feels like there is a lot of gaps in vocals

[analog obsidian]

if you change the vocoder you'll be training from scratch, which needs at least 44 hours of data to provide something good

[oak edge]

aww, I can go online and find the person's 44 hours of concert

[analog obsidian]

it happens with bigger datasets, smaller datasets cant overwrite the pretrain fast enough to experience degradation

[oak edge]

it's going to be really really really hassle though maybe gonna take me 3 days to even get them properly organized

[oak edge]

bigger how many hours are we talking about

[analog obsidian]

i started to notice it when i trained 4-5 hour datasets

and pretrains gets degradation at epoch1

thats how bad the rvc arch is

lol

[oak edge]

ewww

[analog obsidian]

i mean the 48k pretrain of rvc sounds like this

[oak edge]

did you notice or experience similar in contentvec?

[oak edge]

gang I have to ask something to you

[oak edge]

is this epoch 1?

[analog obsidian]

i dont remember how many epochs rvc-boss trained the 48k pretrain

thats the og pretrain 48k btw

[oak edge]

[oak edge]

my epoch 1 - 30 always sounds like this

[analog obsidian]

models do inherit the pretrain degradation too

[oak edge]

like i've trained multiple multiple times and my first upto like 30 epochs sounds exactly like the degradation wav you've sent

it only gets better after reaching about 50? then slowly noticable improvements till like 100-150

then the improvement is kinda not noticable at all

[oak edge]

😭 so that's why all of the contentvec interference I've done had this effect at the beginning?

[analog obsidian]

is not the embedder who is causing the degradation, is something else we dont know lol

the generator just fails

[oak edge]

😭 then how does it improve as we move over to like 50 epochs

[analog obsidian]

once the model has degradation

its ded

[oak edge]

😭 these terms are honestly confusing man

[hot lagoon]

i cant find Gabox's voc_fv4 on mvsep

[oak edge]

if degradation is this thing as you've sent in these attachments, it has been always there in starting epochs, or by degradation are you hinting towards the quality starts to feel like the first attachment after a huge amount of epoch?

https://colab.research.google.com/github/Y-Kitaro/DDSP-SVC_Colaboratory/blob/master/ddsp-svc.ipynb

holy hell

[analog obsidian]

well i call it 'degradation' but honestly doesn't have a name
this 'issue' is when the generator can't average properly the dataset

so you get a mess

[oak edge]

🥹 and are you hinting it can happen both at starting and when overtrained?

[analog obsidian]

basically yeah

[oak edge]

[somber oriole]

why are my downloads failing on chrome ive tried like 30 times and they failed every time my wifi is good i even used ethernet

[analog obsidian]

is not there, but it's in a colab and uvr local

[oak edge]

also doesn't ddsp-svc has some sort of UI? or I have to do everything manually?

[analog obsidian]

https://colab.research.google.com/github/jarredou/Music-Source-Separation-Training-Colab-Inference/blob/main/Music_Source_Separation_Training_(Colab_Inference).ipynb

fv4 is there ^

[oak edge]

@analog obsidian ohhh also you got the link for new kaggle gradio combo?

[analog obsidian]

everything manually yes

[analog obsidian]

nop sorry idk

[oak edge]

awww fuck no i will have to learn a lot

it will be annoying till i get used to it

[oak edge]

oops nvm

let me go to my good ol kaggle and train with contentvec crepe

[viral mason]

crepe? nobody uses that anymore tho

[analog obsidian]

crepe is not viable because it gives a f0 to noise

[oak edge]

why not?

[oak edge]

f0 is just pitch extraction process right?

[analog obsidian]

the second step is calling pitch extraction
thats done using a f0 estimator
the job of this f0 is to teach the model how your voice sounds like at different pitch

[oak edge]

aww fk i have to make it quick

[analog obsidian]

so obviously you dont want the model to learn the f0 of noise

lol

[oak edge]

but what if my data set has absolutely 0 noise?

and I've properly preprocessed using noise gate and truncated?

[analog obsidian]

then rmvpe is still more accurate than crepe for speech

[oak edge]

for songs?

[analog obsidian]

songs too

[oak edge]

😭

[analog obsidian]

rmvpe was made with singing in mind

it was trained with a singing dataset

m4singer

[oak edge]

😭

[analog obsidian]

crepe used speech only

[oak edge]

damn so i'm stuck with rmvpe?

[analog obsidian]

rmvpe is probably the best f0 ever made

not gonna lie

lol

for human voices

[oak edge]

is this rmvpe propoganda

[analog obsidian]

for instruments, etc crepe is better

[oak edge]

alright imma take 1 copy of the song I want with rmvpe (as I've already trained in it)

[analog obsidian]

no big surprise the f0 made for rvc is the best for rvc, crazy

made by members of the rvc team

who had in consideration how rvc works

[oak edge]

💀 rmvpe is solely made by rvc team!?

[analog obsidian]

[oak edge]

wait then what are these benchmarks showing

https://github.com/lars76/pitch-benchmark

[analog obsidian]

instruments , sounds, and speech

f0s are trained using different types of datasets actually

lols

its crazy

[oak edge]

absoluetly no use then to train with crepe?

[analog obsidian]

the f0 doesnt have an impact in quality

it just tells the model how your voice sounds like at different notes

[oak edge]

imma go brute force

[somber oriole]

does anyone have rtx 5000 series and the downloads just wont work at all

bro theres no way the mods are this trash at their job

ive spent 8 hours in here tryna figure out how to get thia done and they are ghosting

[night merlin]

May I ask if anyone knows whether an RVC model trained locally performs better than one trained on Weight?
Or are the results about the same?

My English is poor

[quasi condor]

depends on the gpu they used on weights or ur gpu locally

[knotty moth]

@lost temple I've rejected your dm, please ask here or post a thread in https://discord.com/channels/1159260121998827560/1192011222023950368

[night merlin]

Which element of the GPU determines the quality of the model?
If I want to perform local training, where should I look for recommended GPUs? I am curious that the GPU Applio use is recommended or not, I also wonder if there is any method to know the GPU that Weight use.

Thank you!

[low shard]

!howtoask

[patent trellis]

❓ How to Ask for Help

✅ Before You Ask!

1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.

📝 How to ask?

Tell your:

• Full GPU Name: (e.g., NVIDIA RTX 3060)
• Operating System: (e.g., Windows 11)
• Detailed Description: What were you trying to do and what went wrong?
• Tutorial Used: Link to the guide you were following.
• Screenshot: A picture of the full error message is very helpful.

🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

• (E girl, as an example) catfishing/trolling, scamming, impersonation.
• NSFW/Porn.
• Any illegal activities.
  Requests for these topics will be ignored and may result in moderation action.

<:matsuripray:1159685390156967936> Community Expectations

• Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
• English Only: Please keep all conversations in English.

[low shard]

check https://docs.aihub.gg

[low shard]

that seems like old original wokada, dont use youtube tutorials

[low shard]

I think you just said in dms that you fixed it rn, and yeah i have some irl stuff or sleep sometimes

[brave umbra]

I want to make an AI cover but I couldn't understand it.

[torpid granite]

that's alright now i understand

[torpid granite]

what u using

[brave umbra]

nothing but I want to use phyton and rvs but I dont understand

It is so difficult for me

[brave umbra]

I asked chatgpt and python downloaded a few things. It confused me even more.

[low shard]

vac lite is suggested instead of voicemeeter

[low shard]

please elaborate more

!howtoask

[patent trellis]

❓ How to Ask for Help

✅ Before You Ask!

1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.

📝 How to ask?

Tell your:

• Full GPU Name: (e.g., NVIDIA RTX 3060)
• Operating System: (e.g., Windows 11)
• Detailed Description: What were you trying to do and what went wrong?
• Tutorial Used: Link to the guide you were following.
• Screenshot: A picture of the full error message is very helpful.

🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

• (E girl, as an example) catfishing/trolling, scamming, impersonation.
• NSFW/Porn.
• Any illegal activities.
  Requests for these topics will be ignored and may result in moderation action.

<:matsuripray:1159685390156967936> Community Expectations

• Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
• English Only: Please keep all conversations in English.

[low shard]

damn where are u from? did the support say anything specific?

[low shard]

you aren't getting ghosted, Helpers here are only volunteers, there are very few helpers that help with different things, and ofc they can be busy too

[low shard]

can you tell your pc gpu and operating system?

[brave umbra]

nvidia 5070 and 32ram intel i9 ultra

[low shard]

that's great, check https://docs.aihub.gg/rvc/local/applio/

dont use chatgpt for this

[somber oriole]

nope its the one linked in the guide

[hallow thistle]

"5 hours" runtime is a bit higher than average free runtime which is usually "4 hours" for free users.

[oak edge]

Sri Lanka man I'm not surprised tbh

[low shard]

Nope the docs have no realtime voice changer that has that version, which seems to be original wokada

Either you followed a YouTube tutorial, or you didn't read the full docs tutorial and just clicked the first GitHub link which was just for introduction and proving that it's open source

Anyways what's the issue?

btw I know the issue is solved, but have you seen https://github.com/gradio-app/trackio ?

[lethal shale]

Heyyy I need help again can you help me

[low shard]

Sure if you elaborate :)

[lethal shale]

until yesterday that used to work

i mean until today

now Line1 reacts to nothing I guess

Idk it doesnt work at all

[low shard]

wait weren't you using vonovox or atleast i suggested you that instead the last time we talked?

[lethal shale]

yess i used but I liked that one better

but now it doesnt work

oh wait

it is not windows 10 compatible now?

i remember something like that

[low shard]

why would you prefer wokada deiteris fork over vonovox? vonovox still gets updates

[lethal shale]

i have both but this works better for me

worked I mean

[low shard]

is the input in discord set as line 1?

[lethal shale]

yeess

lo

no

I mean

Output is Line 1 like I sent

[low shard]

i don't think that should be the issue, even tho windows 10 isn't going to receive free updates anymore (unless you're in the UE)

[low shard]

your audio setup should look like:

• wokada deiteris fork:

  • input: microphone
  • output: line 1

• other programs:

  • input: line 1
  • output: headphones

[lethal shale]

of course I have no problem with that

it is Okada not changing the voice

either doesnt take the input

or give the input

Weird thing is

[low shard]

you need to set the audio routting as i said, it will be different on wokada deiteris fork compared to the other programs

[lethal shale]

Line 1 Suddenly got in top in inputs

my default came to fifth

[lethal shale]

I do that

there is no voice in line1

[low shard]

so, you confirm me you did the audio setup just like in #✨│ai-help message ?

[lethal shale]

yup

it is just either Okada doesnt take my input voice or doesn't process

oh

okay

Solved it

[hallow thistle]

The CPU name is either be Intel Core Ultra 9 or an older Intel Core i9, since the letter "i" has been dropped for most recent Intel Core CPUs, they now named "Intel Core 3/5/7" and "Intel Core Ultra 3/5/7/9".

[lethal shale]

I used Firefox

Instead of Chrome

[low shard]

yeah some browsers could have issues some times

[lethal shale]

thanks for your time buddy

have a good day

[low shard]

you too

[hazy remnant]

Hello, 👋 I'm new here on the discord and I'm a total noob to the whole AI thing, I'm playing around with it for the past couple of weeks and got myself set up a SillyTavern client lately, it works pretty well so far for text generation with a model that I run through kobaldcpp but I wanted to try to add text-to-speech to it aswell.
I have installed a TTS Provider which is called SileroTTS and run it locally in a conda environment, just running this combination (kobaldcpp+sillytavern+silero) works totally fine and I can generate tts replies from the chat messages.

Now to my current problem: when I activate the RVC extension inside SillyTavern and try to have a custom voice for the tts then I get two error messages which are these:

1. TypeError: Failed to execute 'blob' on 'Response': body stream already read
2. TypeError: <RVC module> RVC Voice Conversion Failed INTERNAL SERVER ERROR

I hope someone here is maybe able to help and if this is absolutely the wrong place to ask about this topic then I'd be happy about some suggestions in which places to ask for some advice.

[somber oriole]

the link you sent it worked for the download but when i open the file or wtv its stuck in command prompt or it fails instead.

this is what it says

An error occurred connecting to Discord: Could not find Discord installed and running on this machine.

• Running on local URL:

To create a public link, set share=True in launch().

[finite chasm]

yo

[somber oriole]

yo

[minor isle]

anyone has any female models that can laugh?

[nocturne mural]

It just seems like an alternative to wandb and tensorboard, but that would mean changing the code, and for what? what we have is more than enough

Although I think you meant that we could take the code from this and integrate it into Applio’s interface, right?

https://github.com/gradio-app/trackio/blob/main/trackio/ui/main.py

[low shard]

i mean yeah, applio could maybe integrate the tensorboard / wandboard / trackio directly into applio, that's a good idea

[ember fiber]

If I want to cut out background noise, do I have to do it manually or is there a more effective method? i have tried uvr5 instrumental and de noise, but its still there. I have cut out keyboard noises and such by hand but that takes forever. im new to this

[viral mason]

just use gabox's fv4

https://huggingface.co/spaces/TheStinger/UVR5_UI
do this and then just put the vocals of whoever you're making a model of into the slot and use MelBand Roformer | De-Reverb by anvuew, and then finally UVR-De-Echo-Normal.pth in the VR ARCH section (denoise is optional)

[keen pier]

How to make a ai cover

[viral mason]

btw if the anvuew model destroys a voice entirely like this just use de-echo de-reverb and skip the other echo model I suggested there unless there is extra echo

[torpid granite]

I really like de-reveb v2 by suicidal great reverb remover

[viral mason]

idk if it's better than the one by anvuew tho

could test

[kindred sonnet]

How to get the "start_http.bat"

[oak edge]

@analog obsidian i'm back

i trained with crepe

for 200 epochs

[kindred sonnet]

Would anyone be willing to join vc and help me out 😭

[analog obsidian]

soo
what

[oak edge]

it went hell, rmvpe is much better

especially for the parts where the singer screams

like crepe messes it up like the one you shown in degradation

[analog obsidian]

didnt i told u like 3 times to use rmvpe

[cedar pasture]

Hi guys, does anyone know of any models like veo3 where I can generate voice and video, but for longer than 8 seconds?

[simple ore]

without sound, there are plenty

framepack is basically unlimited length

[cedar pasture]

I'll have a look into that ty how about with synchronised audio?

[viral mason]

Download Vonovox 🐠🫧

[magic mango]

Can somone help me to fix the voice changer

[viral mason]

Sure, what gpu do u have and where did u get the download?

[magic mango]

I got it from a video

[viral mason]

Yeah it's most likely outdated like over a year old, even if the video is a few months old

[magic mango]

Oh its says over a year

[viral mason]

They use outdated stuff because they don't know about the good up to date stuff

[magic mango]

where do i go then?

to get the new

[viral mason]

Well what gpu do u have

[magic mango]

How do i see that?

[viral mason]

Open your task manager and click on performance

[magic mango]

AMD RYSEN 5 5600H

Thats my gpu

Hello sir?

Are you still here

[viral mason]

Sorry I was eating I'm back

[magic mango]

Sorry for spamming

[viral mason]

I think that's an ok gpu for realtime

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options 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. GUIDE

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

⚔️ Wokada Deiteris Fork vs Vonovox

For Windows Nvidia, Both Wokada Deiteris fork and Vonovox have similar performance & quality. Users should read the pros and cons for both and choose based on their differences, such as UI and Vonovox's paid effects.
Read Wokada Deiteris Fork Pros&Cons & Vonovox Pros&Cons

⛔ 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]

Download wokada deiteris fork

Read the guide and download it, download link is a little down just gotta scroll a bit

Should say amd

[magic mango]

Here?

[viral mason]

Yes just scroll down a bit until u see the download links

[magic mango]

Here?

[viral mason]

Once u download it unzip the zip file and open up the mmvc server sio exr

What the hell is this

Oh clash royal

[magic mango]

i wish i could screen share it would be easy for me

Im not so good

[viral mason]

That's for a virtual audio cable

Just keep scrolling a bit more

[magic mango]

could we call so i can screen share for you

its easy for me

Please could we do that

Here?

Hello sir

Are you here?

[viral mason]

Yup!

[magic mango]

could i screen share for you?

and you help me like that

[viral mason]

Not rn I really need to use the bathroom lol

So give me a sec

[magic mango]

okay tell me when i can screen share for you

[viral mason]

Mk

[magic mango]

Thank you so much for helping me

I am so happy

[viral mason]

It's no problem I love helping people out with the voice changer

Btw you're not trying to do any e-girl stuff right..m

If u are you're on your own

[magic mango]

No im not

I promise

I just always wanted a voice changer

But you are gonna help me right?

[viral mason]

Of course

[magic mango]

Okay

But am i gonna screen share for you?

[viral mason]

I'll be ready to help u on a VC soon

[magic mango]

okay

i am going in vc5

go in there when you are ready

[viral mason]

just a sec

[magic mango]

okay

are you coming Soon?

[raven ginkgo]

-colab

[patent trellis]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[raven ginkgo]

are there any good colabs to train models rn?

[viral mason]

both kaggle or lightning ai would be better as they give more time for free users

[simple cloak]

Hey

I wanted to ask what programs can u use for Ai pictures

[somber oriole]

does anyone have a rvc voice change for the rtx 5000 series none of the ones i tried work so far

[viral mason]

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Deiteris Fork, with extra features, but supported only for Nvidia GPUs on Windows. and without cloud options 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. GUIDE

• Wokada Deiteris Fork

Most suggested WebUI with the best general support for many platforms. GUIDE

⚔️ Wokada Deiteris Fork vs Vonovox

For Windows Nvidia, Both Wokada Deiteris fork and Vonovox have similar performance & quality. Users should read the pros and cons for both and choose based on their differences, such as UI and Vonovox's paid effects.
Read Wokada Deiteris Fork Pros&Cons & Vonovox Pros&Cons

⛔ 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]

try vonovox

[ember fiber]

Everyone recognizes my voice changer voice because I seem to be speaking too close to my mic. Are there any tricks to make the voice changer more realistic? It's getting annoying. I've tried so many things.

[viral mason]

move away from your mic lol

[ember fiber]

not working

[viral mason]

what voice changer do u use btw and what gpu do u have

[knotty moth]

just admit to them that you're using voice changer

if they're bullying you, it's time to stay away from them

[somber oriole]

u alr told me this

it doesnt work

SOMEONE PLEASEEE GIVE ME A VOICE CHANGER FOR THE rtx 5000 series

[knotty moth]

https://tenor.com/view/dr-evil-how-about-no-gif-13114250

[craggy bough]

it does but you never read the guide

[somber oriole]

dude it literally doesnt i tried it about 20 times

the download fails everytime and i tried 3 different browsers and my wifi is good

its bugged

[somber oriole]

😔😔😔 bru irs been 2 days where the mods at

i think they dead or something

how r they inactive in the server they mod in for this long

[knotty moth]

now pls answer

which voice changer have you tried other than vonovox and the old wokada 18a?

[somber oriole]

there this one someone gave me

i think it was called deteries

and it was a specific part made for 5000 series

it downloaded and all

but when i run it

it says something like set “launch=true” and other stuff

and there is
no launch

i checked all files

[knotty moth]

wdym?

it should never close instantly, so what error message did it show?

[craggy bough]

mods moderate the server

many of us dont use the voice changers

[somber oriole]

it doesnt even open😭

its stuck in command center

and says the thing set true launch

[somber oriole]

ohh ur q mod?

[knotty moth]

that means the terminal window has already shown up but not the gui in the browser

did it also show the link like 127.0.0.1 thingy?

[somber oriole]

yes that it showed that at the top

[somber oriole]

what does that mean my browser didnt change so idk

[knotty moth]

enter the url in the browser

[somber oriole]

no actual way thats how to do it

if it works ima request u to be. amod for this server

[knotty moth]

what

[somber oriole]

im saying no way thats was how to do it all along

im gonna test it rn js opened pc

IT WORKEDDDDD

NO WAYYYYY

THANK YOUUU

now let me see if the voice finally works 6 apps layer

[knotty moth]

ok then

[somber oriole]

i jynzed it i think

it opened the site right away

then it closed and now when i type the url it says error

ok it opened again lms

[knotty moth]

dont hesitate to share the screenshot if something wrong happens

[somber oriole]

ok it opened but im confused on this

it says index and voice model but wont let me drop files

to use my goku voice

[knotty moth]

that is applio realtime

I'd suggest to put the model files in logs folder

[somber oriole]

ok so the whole folder or js the pth and index

[knotty moth]

Applio\logs\yourmodel

[somber oriole]

yes i just put them

[knotty moth]

then click refresh

[somber oriole]

ok i just did

lms

YO IT WORKEDDD

do i press unload vooce

or nah

and for the input do i use regular mic

for some reason i still cant hear my self

testing sound its just 2 ringing noises and gone

[elder harness]

hey ik its probaly me being dumb but im trying to use the voices but its lagging out like crazy. i checked my cpu and it goes to a hundred percent. anyone know if they can help me or got a solution?

[royal kettle]

💀

[somber oriole]

nah im ngl im slow on that i was on the weong spot

but

i still cant hear my self

i turned on monitor sound thing

and i cant hear my self when im using the site

why is that

[royal kettle]

I dont know, as far as i know applio realtime is held together by duck tape

[somber oriole]

what is that

its the only thing i can use since hte others r buggedl

can you tell me as if im not using applio

cause evry site or app i used

never could hear my self

[royal kettle]

Have you ever tried setting the output to line 1 and going into discord and setting your input as line 1 then click test mic?

[somber oriole]

nope never tried let me try rn

bro ain no fkn way

it worked

but one problem the sound is so laggyyy

like my words come out after a minute of saying it

do you know what that is

[elder harness]

im having the same exact problem

[elder harness]

check your chunks

[somber oriole]

oh dang is it also laggy or just delayed

bet lms rn

should i make it higher or lower for best noise

[elder harness]

the lower makes it sound worse but it makes it come quicker

[somber oriole]

ohhhh so which is like the perfect spot u think

or u dk either

[elder harness]

i hear 196 or 256 is a good spot but at the same time idk cause my voice changer doesnt even speak properly rn

[somber oriole]

ok i just put it at 196 ima test it after i try and figure out best settings for goku voice

[elder harness]

ok ok

hey btw what voice changer thing are you using? mine looks way older and diffrent XD

[somber oriole]

im using applio

im just using it cause i got the rtx 5000

[royal kettle]

Just use vonovox

[somber oriole]

it doesnr download

probally bugged

[viral mason]

Your wifi don't like u

[somber oriole]

i used 3 different browsers

nope

[royal kettle]

Wdym

[somber oriole]

not wifi

i downloaded over 20 times and it fails right in the middle

[royal kettle]

vonovox doesnt run on your browser

[somber oriole]

it starts off strong then purposly cancels the download

i tried using

brava

micorsoft edge

and chrome

[royal kettle]

Thats just bad internet

[somber oriole]

ill try again right now and show you

[somber oriole]

i got ethernet bro

[royal kettle]

bad isp

[somber oriole]

its not internet trust me

ill show rn

ok downlaoding rn ill show when it failss

but one thing

do you know why when im testing volume i say hello and it works but really late then when i say a different eord it says hello again from the first time

[royal kettle]

I dont know man, i dont know / use applio realtime

[somber oriole]

im on vonozvoz site rn

where the link to the 5000 series

it was just here

[royal kettle]

Its just setup now

[somber oriole]

?

[royal kettle]

It is only setup now

There is no 50xx version

he combined them

[somber oriole]

oh

will it work

oh betttt

wait

[royal kettle]

Ofc

[somber oriole]

where can i download

[royal kettle]

https://github.com/dr87/Vonovox

[somber oriole]

im ons its

lms rn

do i download both source code zip and tar or just one

im on the 1.6.9

[royal kettle]

just this

[somber oriole]

just downloaded it and extracting rn

one question

why does urs say setup bat

mines never has the bat thing

its always js setup

[royal kettle]

You have extensions hidden

[somber oriole]

how do i unhide lol

ans byw is vonovoz fast and good since u use it

[royal kettle]

https://youtu.be/PoTah9YBG2Y?si=6H6rbu_R12Oh52HB

[somber oriole]

its still downloading but this will be a issue or nah?

[somber oriole]

bet ty

just unhid file

how does vonovoz work lol

[simple ore]

fix that

[somber oriole]

it was alr fixed

i just wanna know how to use this thing

cant hear my self on discord

my volume completely cuts out when i start it why

bru anb can help???

why this gotta be so difficult js for me bro this is crazy why my shit gotta be like this

[craggy bough]

wayyyy too impatient

[somber oriole]

wdym dude u cant be fr

its been 2 days bro

i been asking for help for 2 days and whenever i get to the part that is always bugged evb asleep or sum

[craggy bough]

youre on a server where every helper is a volunteer

you cant expect immediate replies

[somber oriole]

2 days i shouldnt expect a reply?

nah bro idk abt that

can u js help me and get it over with bru plsss

[craggy bough]

and every time someone tries helping you decide to switch to something else

[somber oriole]

why cant i hear my self i have tried 3 different mic testing aites

[somber oriole]

CAUSE THE ONE I USE DONT WORK😭😭😭

and whenever i get to the most important part they ghost for a day or sum

[somber oriole]

i hate my self bru why am i so dumb

YO BFORE U GO

[craggy bough]

just read the docs

dont be lazy

[somber oriole]

just one thing bro why is it so delayed and cuts out at times

its like rly delayed

im using vonovoz

yk what i could change to fix it?

[craggy bough]

no idea

how many times do i have to tell you that i dont use it

[somber oriole]

u never said that

btw does ur w okada work good or no?

[craggy bough]

wokada is a voice changer

[somber oriole]

yes ik does it woek good like no delays and stuff?

[craggy bough]

this guy

can you read

holy

[somber oriole]

ohhhhhhhhhh

[craggy bough]

do you need me to type in larger font?

[somber oriole]

im slow bru

lolllllnah bro its cool

😭

mb thanks tho bro

[keen escarp]

a

does anyone know why the Virtual cable audio doesnt register my voice

on mac !

[marble forge]

Anyone know of the best locally run (runnable on my own pc -- I have a 3060) ai song generator (or a song cover maker, or even one that allows me to use a rvc voice model to make a song with that voice). I am wanting to use one of these ai programs to make a "unreleased" song of an artist.

[elfin igloo]

Does this not work anymore? https://huggingface.co/wok000/vcclient000/resolve/main/vcclient_win_cuda_2.0.65-beta.zip?download=true

It won't download'

Where do I download?

[low shard]

that’s an old version of the beta of original wokada

which isn’t suggested

don’t use youtube/video tutorials ofr ai realtime voice changers

!howtoask

[patent trellis]

❓ How to Ask for Help

✅ Before You Ask!

1. Check Docs & Guides: Your answer may already be in the AI Hub Docs or the https://discord.com/channels/1159260121998827560/1159513888199540817 channel.
2. Search the https://discord.com/channels/1159260121998827560/1192011222023950368 : Look for existing posts that solve your issue. Do not invade someone else's post.

📝 How to ask?

Tell your:

• Full GPU Name: (e.g., NVIDIA RTX 3060)
• Operating System: (e.g., Windows 11)
• Detailed Description: What were you trying to do and what went wrong?
• Tutorial Used: Link to the guide you were following.
• Screenshot: A picture of the full error message is very helpful.

🚫 Prohibited Topics (We Will NOT Help With These)

To maintain a legal, safe & ethical community, we will NOT provide help for:

• (E girl, as an example) catfishing/trolling, scamming, impersonation.
• NSFW/Porn.
• Any illegal activities.
  Requests for these topics will be ignored and may result in moderation action.

<:matsuripray:1159685390156967936> Community Expectations

• Be Polite & Patient: Our helpers are volunteers. You may ping the Helpers role once.
• English Only: Please keep all conversations in English.

[elfin igloo]

Is there one similar that works the same?

[low shard]

please read the guidelines above and elaborate

[elfin igloo]

ok

I still don't know how to download

I liked that beta one

It is simple

[low shard]

Read the help guidelines, it tells you what you need to elaborate to make us help you

[low shard]

it’s not suggested

[oak edge]

yoooo @low shard

do you know hwo to use formanting?

[low shard]

are you talking about the formant shift in vonovox?

[oak edge]

no in rvc

[oak edge]

i have a male voice I want to swap it with female song

well the pronounciation is really weird for starters, like idk how to explain it, the words are correct but the way is really weird

[low shard]

it’s better to lower the pitch and play around with it

[oak edge]

the pitch is correct as I've turned it down to -12

but the pronounciation is awkward af

like there is a feminine touch in the pronounciation?

so i need some help in figuring out what does timbre and quefrency means

[magic mango]

Could i get some help?

[knotty moth]

!howtoask