✨│ai-help

[sand forge]

thats my mic

right

its done?

[prime badge]

With Deiteris' W Okada Fork or Vonovox is there any way to slow down the voice?

[viral mason]

@simple ore I have this error in the kaggle applio space, is there a new link or method to get it working

currently using the 3.4.0 branch

I remember someone saying to update applio but I don't remember who

[simple ore]

looks like you cloned the latest code?

not 3.4.0 branch

[viral mason]

I have, I've used the method you said to use

it's the third option right?

I've tried it twice now and it gives that error at the third cell

[simple ore]

this seems right

and should be cloning the branch without realtime

[viral mason]

which is that

[simple ore]

anyway, make a new cell

paste !sudo apt-get update && sudo apt-get install -y python3.11-dev portaudio19-dev then run it

that should fix the error with portaudio

[viral mason]

should that cell be before the first second or third cell

or does the order not matter with that

[shrewd jay]

What rvc should I use?

[viral mason]

I really hope I don't have to do this every time

[simple ore]

make sure the notebook clones 3.4.0

[viral mason]

by not deleting that part of the code in the first cell right

[simple ore]

yes

[shrewd jay]

Link?

[simple ore]

that's not for you

[viral mason]

bro probably replied to the wrong message

[viral mason]

thx it worked, if I need to do this again in the future is there a way to easily upload the version of the notebook onto github and just find it the same way I do the usual kaggle applio notebook for importing it

[shrewd jay]

Yall got any good Vonovox settings..?

[simple ore]

you only need to import the applio notebook once

(until we switch the version)

it stays unchanged in Kaggle in your account and you dont need to edit it

[viral mason]

I delete the notebook after it's full, as idk how to remove those files

can't keep training once it's full

[simple ore]

can probably nuke the model name in the logs

[viral mason]

how?

only option is for downloading

unless it's something to do with code

[whole mica]

I am looking for an AI program that can convert my 3D renderings into an icon style while still making them look like my 3D model. I tried Midjourney, but when it comes to icons, I can't get anything decent out of it. ChatGPT comes closer, but I have over 100 icons to generate, and at some point ChatGPT always deviates from the style, usually after 2-3 generations. Do you know of any software that can consistently generate the same style for me? Where I can specify a style with images and my 3D model, and then it is converted for me?

[half phoenix]

i need a rly good vc for a girl im jus gonna troll to my frnds (ive got a rly bad accent )

oh im sorry i didnt read the rules sorry

np

[viral mason]

The number of these people grows each day

[half phoenix]

im sorry twin 😭 forgive me

[rapid bison]

uhh i need help

i got error when i launch voice changer

i wanted to send a photo

butt

okeey sir

[tropic mountain]

my voicechanger cuts out while im talking how do i fix it

[fluid lion]

How much "faster" would a 5090 be vs a 4090 for rvc? Would it be a noticeable difference in ping/delay i can run it at

[viral mason]

it's ok man, as long as you don't do it again

[simple ore]

using file manager URL?

using training UI at the bottom

[viral mason]

I don't know what that means

[devout bone]

For some reason I have all the input and output right but I still can’t hear the voice

[nocturne mesa]

I can't hear my mic

Like my mic doesn't work

when i use the virtual audio cable

[south verge]

hey how can i make the delay a bit shorter?

[tawdry needle]

hello does anyone know how to make rvc work on 50 series ?

[plain zinc]

Do you know why it doesn't detect my voice or why I can't hear it? Yesterday the program was fine.

[strange wraith]

is there a video tut ?

[rancid pine]

Are these normal graphs while training or should I restart altogether?

[simple ore]

you did not stop the training cleanly at the end a saved epoch, so the chart have a step back

look at avg 50 charts instead

[oak edge]

oiii @simple ore

i'm training a 1hours 40 mins data set, it's been 345 epochs Y_Y

how much can I push more

[simple ore]

it is likely done already

check the charts

test the saved models

[oak edge]

i actually don't know how to check properly, is loss/g/total smoothed to 0.987 good to check?

[simple ore]

click 'Scalars' at the top

then expand two avg_50 sections and collapse others

[oak edge]

it's like this

[simple ore]

what abouy loss_avg_50?

[oak edge]

lemme see

[oak edge]

there are a lot which one you want

[simple ore]

other 3 charts?

[oak edge]

lemme send

[simple ore]

something happened after 30k steps

probably not worth looking at those

so take the model saved around that step and give it a test

fp16 is not quite stable and it exploded for you

you can probably use a much smaller dataset.. 30-40 minutes

[oak edge]

i trained it different times

using gdrive backup

[simple ore]

something got messed up

[oak edge]

how bad

[simple ore]

perhaps the precision got switched mid-way

[oak edge]

is that a good thing or a bad thing

[simple ore]

like you started training with applio 3.3.0, then the discriminator got fixed in 3.4.0 and default precision was set to fp16

in this case you need to restart

sorry about that, you better off using 3.4.0 anyway

[oak edge]

💀 i've been trainin for 15 days

[simple ore]

but give the model from ~30k steps a try

[oak edge]

each session gives me around 20 epochs

[oak edge]

can I know did the quality have gotten worse after 30k

cuz im not sure what you mean by that 30k, and what's before and after 30k

[simple ore]

the model name has a number of steps

like

your model got messed up after ~35k steps

most likely you did train at fp32 at the start, then we had 3.4.0 released and you resumed with fp16

then you also resumed after 58k steps with a wrong batch size

[low shard]

please elaborate, this is a general AI server

[low shard]

what program are you talking about? please elaborate more

[low shard]

RVC does work on rtx 50 series, what program and tutorial link are you using?

[low shard]

hello, could you please elaborate more on what you're talking about?

!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 lega, 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.

[south verge]

As in the real to voice it has almost a 6-10 second delay and I was wondering if there’s anyway to shorten that delay

[low shard]

This is a General AI Server, please elaborate more your PC specs, the tutorial you're using and a screenshot of the settings, we unfortunately can't know which program and settings you're talking about (I'm guessing a realtime voice changer, but there are different programs and versions)

[oak edge]

AAHH YES I CHANGED BATCH SIZE

[livid hollow]

can anyone solve it :--

---

cpt = torch.load(modelPath, map_location="cpu")
2025-09-08 11:17:20,560 ERROR [MMVC_Rest_Fileuploader] 'config'
Traceback (most recent call last):
File "restapi/MMVC_Rest_Fileuploader.py", line 67, in post_load_model
File "voice_changer/VoiceChangerManager.py", line 118, in load_model
slotInfo = RVCModelSlotGenerator.load_model(params)
File "voice_changer/RVC/RVCModelSlotGenerator.py", line 42, in load_model
slotInfo = cls._setInfoByPytorch(modelPath, slotInfo)
File "voice_changer/RVC/RVCModelSlotGenerator.py", line 58, in _setInfoByPytorch
config_len = len(cpt["config"])
KeyError: 'config'

[simple ore]

you're trying to use a non RVC model

[livid hollow]

Please, can you share the correct link for the RVC model I can upload?

[simple ore]

whatever you've downloaded is probably some old SVC model or something

how would I know that?

its like you went to a junk yard, ripped out a carburator from a 1960s car and now trying to plug it into a modern car with fuel injectors

[livid hollow]

I just want to use basic voice conversion, like any voice.

[hallow thistle]

I just found out on Saturday that you can listen to your "microphone" or "Line 1 (Virtual Audio Cable)" on VLC. Can be useful if you wanna hear what "W-Okada with output set to Line 1" is outputting, although the VLC can add up the delay for the audio you selected to hear but not really affect its actual "input" signal overall.

[hallow thistle]

!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 lega, 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.

[hallow thistle]

That one from other guy was being sarcastic for sure. SVC and RVC are two whole different voice model architectures. SVC (so-vits-svc) is the old one from 2023, while RVC (Retrieval-based Voice Conversion) is the newer one and currently widely used. Also, SVC voice model cannot be used in an RVC program.

[livid hollow]

apologies..

I'm using the link below and following every step in it in kaggle,

When I finished uploading the model i got some errors related to config in kaggle...

https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-kaggle/

[hallow thistle]

[livid hollow]

[prime badge]

I am using RVC b2332 Nvidia-Cuda, how do you slow the voice down? When I speak i can hear it in the monitor speaking faster then what I speak

[hallow thistle]

In this context, RVC voice model can only be used in Deiteris fork W-Okada, even as locally and online, because any voice model other than RVC was removed from that W-Okada.

[hallow thistle]

Why slow? Do you mean like to make the audio delay more or something?

[marble cedar]

check your i/o sample rates

[prime badge]

Not delay from when I start to speak but slow down the speaking between words

[livid hollow]

can you share any links from where i can download and use models for W-okada (kaggle)

[hallow thistle]

!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 lega, 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.

[hallow thistle]

What do you use W-Okada for? Funny or trolling or catfishing? And what is your PC GPU?

[hallow thistle]

Huh? Can you explain that again?

[prime badge]

If I say a sentence in my normal speaking speed I can hear it in the monitor speaking faster then I speak which leads it to sound weird

[livid hollow]

my goal is to use it on call centers... I've tried elevnlabs but there's no real time voice conversation option....

and its for one of my projects and not for catfishing or other spammy thing

[prime badge]

I find myself have to purposely speak slower

[livid hollow]

my end goal is that multiple people can use it so I'm only thinking about production grade solutions, so basically runpod or other hosting things

[hallow thistle]

Why would you use it for "call center"?

[livid hollow]

okay, mind if i dm and explain my needs?

[marble cedar]

Does anyone know the best set of resources or guides for hosting a huge model on a Google Cloud Compute Engine VM with a T4 GPU

[hallow thistle]

If nothing is personal, no need to direct message me.

[livid hollow]

try runpod maybe

[livid hollow]

okay so it's a project for one of my clients.

[hallow thistle]

Nah, you can't make W-Okada to slow itself down. Instead, there can be issues with your W-Okada settings, the program lagging or your PC lagging.

[hallow thistle]

What is your client's PC GPU?

[livid hollow]

As I’ve said, I’m using a cloud-based GPU, so that’s not a problem. Currently, I’m on the Kaggle free tier (T4) and at least I’ve got an interface for real-time, which I wasn’t getting with direct RunPod GPUs. But with Deiteris’ W Okada Fork on Kaggle it worked — the problem I’m facing now is a KeyError: ‘config’

[hallow thistle]

I've never used Runpod service for myself. I only use Google Colab and Kaggle for most parts. But how did you integrate the whole "Deiteris fork W-Okada" into Runpod by yourself? And yes, there are download links for fork W-Okada.

[balmy tulip]

if im using kaggle for w-okada do i have to rerun the cells every time i want to use the voice changer?

[livid hollow]

Not the fork version — I was using the old RVC repo which is discontinued. With that, I could only do basic voice conversion (uploading audio and getting output), not real-time.

Yesterday I found out about this server and tried the W-Okada fork, which surprisingly looks like it supports real-time.

But now I’m stuck on one error… someone mentioned I’m using RVC-based models, but I’m not sure what other types of models I should actually be using?

[hallow thistle]

Yes, you'd have to run each cell after you start up the notebook, but just to ensure that the notebook should work as intended.

[hallow thistle]

Deiteris' fork W-Okada, as what I said, only supports RVC voice models.

[livid hollow]

select run all every time you refresh the page

[hallow thistle]

These are RVC voice models. Their file extension is .pth, and its file size should always around 55MB. Most of RVC voice models often come with an index file, which is where it stores accent of that voice model.

[livid hollow]

can i use models from the link below ?

https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main

[hallow thistle]

You can't use "pretrained voice models" that come from any RVC repository. These pth files are too large to be used as normal RVC voice model, and they all supposed to be used in training another voice model.

Instead, you might wanna find "voice models" in #1175430844685484042. The thing is there is no known "generic" voice model there; most of which are of characters and famous people.

Typical pth file size for a normal RVC voice model is around 55MB; any size that greater (> 60MB) or less (< 40MB) than this doesn't sound right.

[livid hollow]

well well well that's some help I was expecting.
thank you very much WBN 🙏

[livid hollow]

one more question will i also get an index file from here and do i need to upload that too in the interface for real-time convo ?

[hallow thistle]

Most RVC voice models typically come with an index file. You can use both files in an RVC program. But in W-Okada, the index file is usually rarely used anyways, so only pth file needed.

[livid hollow]

great 👍🏻 appreciate your response.

[hallow thistle]

Thank you, and have a good day.

[tawdry needle]

i used the one that you sent in the dev channel but it is a web version and it only works on server and not client for me and even after i make it work it has many problems like not beign realstic or glitching a lot while breathing or luaghing i also sent u a dm

[low shard]

I'm not sure what you're talking about, please elaborate more on what you mean, remember that RVC means Retrieval-based-Voice-Conversion, not realtime voice changer

[tawdry needle]

https://github.com/tg-develop/voice-changer/releases/tag/b2364

i used this one for nvidia

i have a 5070

it opens in browser and my mic sound only work in it when i pick the server option with windows direct

client doesn't work

[low shard]

oh that's not rvc, that's a modified version of the wokada deiteris fork, it's meant to work in realtime for roleplay

[tawdry needle]

and even after it work it has too many glitchs

[tawdry needle]

then what is the best version i could use ?

[low shard]

it's normal that it opens in ur browser, try different browsers and be sure to give microphone perms, then show a screenshot of ur settings

[low shard]

This is a General AI Server, the program you need depends on what you want to do

RVC = Retrieval-based-Voice-Conversion, the best Few Shots Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models. Technically, Mainline RVC does have a go-realtime.bat (aka RVC-GUI), but it's pretty messy and outdated so it's extremely not suggested for realtime. There also updated forks with extra features like Applio.

Wokada = uses RVC for realtime inference.

Vonovox = Another Realtime Voice Changer based on RVC, with similar quality and performance to wokada deiteris fork but other perks

Wokada isn't RVC

and RVC isnt realtime voice changing

[tawdry needle]

i sent you the pictures in dms

[tawdry needle]

i need it for vc in games , streaming etc

[low shard]

Oh I didn't check you dmed me, that's because it was in a request lol, i will reply there

[tawdry needle]

oki

[knotty moth]

please check settings for the client mode

you can first tell me if you need to show the screenshot

[livid hollow]

Wokada or Vonovox which one is best for real-time according to you ?

[low shard]

there isn't really a "best" one, there are different Wokada Versions/Forks (Original, Deiteris, Tg-Develop's one), and each has it's own pros&cons, it also depends on your specs since vonovox works only nvidia users on windows (as of now)

[tawdry needle]

sure look dms i am just trying to figure out models and stuff

[plain zinc]

It was about RVC, but I was able to solve it, it only took me 3 hours

[viral mason]

rvc is many different things lol

[compact sorrel]

anyone know why wokada gives me over 11000ms ?

[low shard]

if you're playing a game, play on lowest graphics 1080p 60fps cap

show a screenshot of your settings, elaborate more since there are different versions

[compact sorrel]

sure can i get pic perms?

[low shard]

!give-media-perms 1h @compact sorrel

Done!

[compact sorrel]

capped myself on 60 fps roblox and still 17000 ms

[low shard]

from the interface, that looks like a 2 year old outdated original wokada, did you use a youtube video tutorial?

[compact sorrel]

yep

could u provide me a link to a newer version?

[low shard]

yep, don't use video tutorials for realtime voice changers, they are pretty old

delete the folder

also, if you have vb audio cable

[compact sorrel]

ok

[compact sorrel]

i do

[low shard]

uninstall it from windows app settings

[compact sorrel]

kk

[low shard]

sure, could you tell me your pc gpu? because there are different versions depending on that

[compact sorrel]

amd

[low shard]

I'm guessing you're on windows 10/11

[compact sorrel]

11

[low shard]

could you tell the full gpu name please?

to check if it's also good enough

[compact sorrel]

is that good enough?

[low shard]

should be good enough

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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

[low shard]

you can use either wokada deiteris fork, or tg-develop's fork (which is just a deiteris fork but with some Qaulity Of Life additions, don't expect too much from it ofc)
You can check each pros&cons in the docs :D

[craggy dune]

Just wondering, im testing some models and even though i have the monitor set to none, i can still hear myself play back. How do you turn that off?

[low shard]

Could you please show a screenshot of your settings? I?m guessing you're talking about realtime voice changer, this is a general ai server, so we don't know which program you might be using

!give-media-perms 1h @craggy dune

[compact sorrel]

okay so im on https://github.com/deiteris/voice-changer and im confused where and which file do i download ?

[craggy dune]

its realtime voice changer by okada

[low shard]

always check the Triangle, it's very important for non nvidia gpus
turn on echo, lower your headphones volume and keep your microphone far

[low shard]

Note that those 3 links are just for reference to the Source Code Github Repositories of both projects, you should instead follow the guide below.
Please read the guide below

[craggy dune]

okay, well i've done that but as i understand, having monitor selected give you audio feedback

but i have nothing selected for monitor and i can hear myself through my headphones

[low shard]

monitor is to hear yourself, since you don't want that and it's not used to your headphones, you seem to have issues with echo,
have you tried what I said?

[craggy dune]

i enabled echo, swich back from cpu to gpu

[low shard]

still having echo issues?

[craggy dune]

well it just sounds like monitor audio, to hear playback

but i dont want that

[low shard]

you aren't using monitor, this isn't related to it, this is related to an issue with echo,
have you tried also lowering ur headphones volume? and are you sure you setup correctly vac lite?

[low shard]

also, be aware that extra over 2.7 could cause cutoff issues in some cases

[craggy dune]

What does that mean, echo

I just dont want to hear myself

and what i was lead to believe monitor lets you hear your audio playback

which is what im hearing

[lethal iron]

Does anyobody have a up to date tutorial how to install comfyui-zluda on My rx 6750xt

[low shard]

echo means the repetition of a sound, this isn't a setting, it's an issue caused by your setup

[craggy dune]

What is vac light

[craggy dune]

The sound isn't repeating like a loop.

When i say hi, i hear hi back with the voice changer once

[low shard]

vac lite, it's the one you're using right now as Line 1, used to get the output of the voice changer as input in other programs
you sure you did this?

can you show a screenshot of the recording and playback tab?

[craggy dune]

Okay i fixed it. I clicked on the clear settings button and now i cant hear myself

perfect

[low shard]

That's pretty weird? Might have been some settings stuck you messed with, since that shouldn't have happened

But glad to hear it's fixed now

[craggy dune]

and also my line 1 cable settings are set up properly

so yeah idk

weird bug

[sand tree]

Hello! Is there anythig I can do about weird pronounciation issues? For example, when i try to say words like "going" it pronounces as "gohee" as if it cant produce the "ing" sound

[craggy dune]

Why shouldn't i do 5 second extra?

[tacit verge]

why my voice lagging and cutting out

[low shard]

5 seconds extra could maybe sound better, but in some cases, any value over 2.7 could cause a bug in some models to cutoff

[low shard]

Hello, could you please elaborate more? This is a general ai server, and there are many different programs, showing a screenshot would deffo help us 😁

[shy gazelle]

why when I turn on the downloaded voice, the res parameter increases to 8000, resulting in a huge delay

[dry coyote]

all i hear in my RVC is mumbling

i use deiteris's form

fork

[low shard]

Hello, could you elaborate more please? Such as your GPU and a screenshot

[viral mason]

All u do is download it, download vac lite, extract both, install vac lite on your PC then restart it (if needed), run the setup file in vonovox, then once that's done run the start file in vonovox

[low shard]

There's no RVC deiteris' fork, rvc doesn't mean realtime voice changer

RVC = Retrieval-based-Voice-Conversion, the best Few Shots Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models. Technically, Mainline RVC does have a go-realtime.bat (aka RVC-GUI), but it's pretty messy and outdated so it's extremely not suggested for realtime. There also updated forks with extra features like Applio.

Wokada = uses RVC for realtime inference. There's 2 main versions, Original made by Wok, and the most suggested one is Deiteris Fork (modified version)

Vonovox = Another Realtime Voice Changer based on RVC, with similar quality and performance to wokada deiteris fork but other perks

Maybe you meant wokada deiteris' fork?

[low shard]

What's your PC GPU and OS?

[knotty moth]

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

for Nvidia gpus only

[low shard]

-# downloads it on Intel Mac

[viral mason]

Some people are bad at reading (lazy) so I explained how to do it without a guide

[knotty moth]

more ppl falling to the random tuts in youtube

[viral mason]

Yeah, I wish those yt tutorials that are several years old would get taken down

Tbh all of them should be taken down especially since a lottttt use them specifically for the one thing (egrill shit) that 85% of the people that join here for want

[wanton rapids]

RTX 4060 Windows 11

[wanton rapids]

I heard something about python, do I need that?

It feels odd that a voice changer is only 6.6 mb

I wouldve used W-Okada but vonovox seems a bit better

[viral mason]

I believe so, I can't fully remember if you do but it's most likely needed just to be easier when it is downloaded and used

[low shard]

great, you will be able to use vonovox then

@viral mason @wanton rapids You don't need to install python externally/globally, it comes bundled with Vonovox

[viral mason]

that information will be helpful with the next time someone asks about that

ty Nick

[wanton rapids]

Ah alright

Thanks

Btw uh theres 3 different setup files in vac lite

Which one do i run

Setup, setup 64 or setup64a

[low shard]

setup64

It's written in the guide, and the next step will be crucial, be sure to follow it :)

[wanton rapids]

Which is?

[wanton rapids]

Ah i see, theres line 1 in both input and output

I switch back to normal?

[low shard]

your recording tab default should be ur usual input device
your playback tab default should be ur usual output device

[wanton rapids]

So yes

Oki

Done

In vonovox do i run setup? The windows batch file

[low shard]

after you extracted it, yup

[wanton rapids]

alr it opened cmd idk when to expect the finish

[low shard]

it will tell you when it finished

[wanton rapids]

Splendid

holy

3.5 gb

yea now it feels real

the models usually sound normal only in english/japanese?

i mean i see that most of them are english/japanese

[ancient trout]

Hi guys I’d like to create an AI that plays perfectly any time of SHOOT EM UP game but idk where to start and know nothing about programming.

Is there any tools that can help me achieve it with no programming skills ?

Thx in advance 🙂

[fallow lava]

Guys, what is the recommended specs for streaming realtime with Deiteris' W Okada Fork?

If I were to shop for a GPU now, which would you recommend?

[low shard]

Do you need any other help?

[wanton rapids]

im confused on how to actually use the voice changer

i just download a model and thats it?

i think i gotta set the virtual cable as the input device actually

[low shard]

upload a model in a slot, use f0 as rmvpe, extra as 3.0, output as line 1, input as microphone and you can start it

[wanton rapids]

thanks alot

[low shard]

you're welcome, let me know if it's solved :)

[wanton rapids]

and uh, if im in call do i have to change output device?

[low shard]

To use the voice changer in other programs, be sure to use those audio routing settings

• vonovox:
  • input: microphone
  • output: Line 1
• other programs:
  • input: line 1
  • output: headphones

[wanton rapids]

ah yea makes sense

vonovox output is set to line 1 just so i can hear the voice changer myself right?

[low shard]

nope, it's used so that the converted voice can be used as input in the other programs

you can't hear yourself in vonovox unless you do a workaround https://docs.aihub.gg/realtime-voice-changer/local/vonovox/#how-can-i-hear-myself

[wanton rapids]

oooooooh

[low shard]

Is there something you need to understand still? I'm here to help ofc

[wanton rapids]

all good, thanks

[rancid pine]

Is contentvec or spin better for non-English RVC models? I'm looking for a Chinese and Japanese pretrained model, please give me recommendations

[simple ore]

there are chinese and japanese embedders, but there are no chinese or japanese rvc pretrains done using them

[rancid pine]

Shucks

[simple ore]

spin-v2 is what we got now

[obtuse kestrel]

hello, do we still need to install this "python 3.12.7" to our windows, i use fork okada

[low shard]

I'm guessing you're talkin about either deiteris or tg-develop forks, since "fork" just means modified version in tech,
You don't need to install any global python, the program comes bundled with it

[wanton rapids]

@low shard sorry if im bothering, do you know any good female voice models? the one i downloaded sounds kind of weird when i use it

[viral mason]

What are you using those for hmmmmmmmmm? 🙂

[wanton rapids]

sshhh

no one has to know

but basically, even the ones meant for english sound like they were made for japanese or smth

idk if im using it wrong

[viral mason]

Do you have an accent by chance?

[wanton rapids]

eehhh a little bit maybe

[viral mason]

The voice could be trying to copy that too

[subtle gate]

how to fix lag and cutting out

[wanton rapids]

a bit of a slavic accent but its not anything crazy

[viral mason]

You could try searching the Russian tag in the voice model section

[subtle gate]

it is over 5 sec

[viral mason]

Might help

[wanton rapids]

im not russian lol, bosnian

like i said, the accent is there but u can barely hear it

i dont think it should present a problem

[viral mason]

Fix your settings

[wanton rapids]

maybe its the pitch?

[viral mason]

My bad lol

[viral mason]

If you're a dude try turning the pitch to 3, 6, if it really needs it out up to maybe 8 or 10

Usually 3 is good

[wanton rapids]

its at 12 rn

[viral mason]

Oop

[wanton rapids]

i have a bit of a deep voice 💔

11.7 to be specific

[viral mason]

Hmm

Just mess with the pitch until it sounds like the model, btw it really helps to actually put effort in maybe making your actual voice a bit higher pitched when talking

[wanton rapids]

i would do that..if i was home alone

[viral mason]

I change my voice a little in how I talk with some models like if I use the battle droid compared to just a normal person

[wanton rapids]

either way i dont think makima is it

💔

[viral mason]

Dang

Well try finding or u could make a model that fits better

[wanton rapids]

yea i need a better voice

[wanton rapids]

you got any good ones?

[viral mason]

Uhhh probably not regular female models but u could check my weights page if u wanna browse through everything

[wanton rapids]

just pick a random one lol

ill try it

[viral mason]

I mean u could look, I kinda have to use the bathroom rn 🙏

[wanton rapids]

ah

how do i find ur page

[obtuse kestrel]

okay thanks

[hexed osprey]

Can someone tell me the setup for wokada with my 3060 ti and AMD Ryzen 5 3600 6 core? thanks

[viral mason]

Just look on weights, and search my name

[low shard]

Sure, are you having any issues or want a guide to install? I'm guessing you're on windows 10 or 11

[timid talon]

can i ask for help?

[lusty whale]

is it just me or did wokada recently start bugging where your not able to hear yourself nor hear anything coming out of your mic to others

[narrow wraith]

How can I get the voice of a character?

[hallow thistle]

#1175430844685484042

[hallow thistle]

!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 lega, 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.

[hallow thistle]

You can ask anything, it doesn't have to be "can I ask a question?". Also, make sure to read help guidelines before start asking.

[short oracle]

hii i would like to ask how to reduce rtc latency

[low shard]

rtc? could you please elaborate more on what you mean / program you're talking about? Since this is a general AI Server, there are multiple programs with different versions based on OS & GPU
I hope you understand :)

[short oracle]

oh uhh lemme see wat the name is

i use deiteris wokada fork

[low shard]

I'm guessing wokada deiteris' fork (modified version) b2332, could u please show a screenshot of ur settings?

!give-media-perms 1h @short oracle

[short oracle]

its loading D:

[low shard]

btw, are u using it in games?

[short oracle]

i am streaming game

while using it

[low shard]

Play on lowest graphics 1080/720p 60/30fps
the gpu prioritizes the game since it's rendering something on the screen, so you have to give more resources to the voice changer

[short oracle]

i have...

5080

[low shard]

lowest graphics 1080p 60fps cap is usually suggested for everyone, because this way it can give more resources to the voice changer to have less delay
The GPU is doing 2 very intensive tasks at the same time, if you get what I mean
I mean you're free to experiment to see, but playing on like 4k max graphics is going to take more resources which can higher up the delay, ofc this can depend by the game too

[low shard]

on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:

• Advanced Settings -> Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
• Advanced Settings -> Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
• Advanced Settings -> Crossfade Lenght: Controls how smoothly the AI stitches different processed parts "chunks" of your voice back together. 0.1 for fastest voice, 0.15 for improved quality but increases delay by ~50 ms
• Reduce the delay on Windows via the Wasapi / Asio Guide

You could also give Vonovox a try, it still gets performance updates, you can cehck its pros&cons if it fits your needs btw

[short oracle]

i mainly wanna use it when hanging out with friends and streaming them game

[short oracle]

i would never doo that, (totaly not streaming metal gear delta on 4k)

[low shard]

Yeah those are optional settings, each is explained to help with delay, like wasapi

They are in the Advanced settings section, and in the docs

[short oracle]

okayy thank youu

[low shard]

Btw vonovox can do that too, it's another realtime voice changer that uses RVC models, but it still gets updates especially about performance, it's in the docs if u wanna check it, your choice

[short oracle]

is it better ?

[low shard]

You can give it a test, there's no official benchmark test that I can give you

[short oracle]

is there a one that i could run on like phone, and just stream the audio to pc?

[low shard]

It will never work on your phone because of lack of power + no VAC

[short oracle]

ahh i see

[agile relic]

I'm having a problem installing it

[agile relic]

okay I fixed it myself

[brittle wing]

hi new here - I'm looking for a voice changer to do agent voiceline soundboard ingame - can anyone tell me which one can I use?

[oak edge]

@low shard

can you explain how these live voice changers work?

[hallow thistle]

Basically, the "realtime voice changer" program converts your voice in realtime with a given voice model.

Similar to RVC that converts an audio with an AI voice model, but this one converts and processes audio in realtime.

[agile relic]

@hallow thistle I'm having trouble getting the cable output VBCABLE to get the voice changer to sound in it and have it working.
Is there any guide link or information available to get this fixed?

I'd appreciate even a copy paste link from a previous question of this type as well

[hallow thistle]

Use Virtual Audio Cable lite instead of VB-Cable. Also, did you follow any tutorial before this?

[agile relic]

I was using some youtuber's tutorial who told to use VB-Cable and the comments on the video really said it was working, I'll try the one you suggested, do I uninstall vb cable?

[hallow thistle]

W-Okada can't work on phone (Android and iOS) like in desktop PCs, as there's no virtual cable for this and the audio system is far different from desktop one.

[agile relic]

I annalyzed in virustotal the (vac) and is this a false positive?

• Win32.Troj.Undef.a

Virtual Audio Cable exe program

[hallow thistle]

Yes, you can uninstall VB-Cable to save up your PC disk space.

[hallow thistle]

Why would you use an antivirus program for this? Some antivirus programs, aside from Microsoft Defender, can give false positives against certain programs.

Try download the program from this link. https://software.muzychenko.net/freeware/vac470lite.zip

And what is your PC GPU?

[agile relic]

I was about to ask if zip or exe, thanks

[agile relic]

can I dm ss

I can get it in text wait

[hallow thistle]

To check your PC GPU, open Task Manager, go to Performance tab and locate GPU 0 or GPU 1.

[agile relic]

System Information
Current Date/Time: Wednesday, September 10, 2025, 3:07:51 PM
Computer Name: DESKTOP-9BIQ8BE
Operating System: Windows 10 Pro 64-bit (10.0, Build 19045)
Language: English (Regional Setting: English)
System Manufacturer: LENOVO
System Model: 82H8
BIOS: GGCN29WW
Processor: 11th Gen Intel(R) Core(TM) i3-1115G4 @ 3.00GHz (4 CPUs), ~3.0GHz
Memory: 8192MB RAM
Page file: 11621MB used, 8657MB available
DirectX Version: DirectX 12

Name: Intel(R) UHD Graphics
Manufacturer: Intel Corporation
Chip Type: Intel(R) UHD Graphics Family
DAC Type: Internal
Device Type: Full Display Device
Approx. Total Memory: 4123 MB
Display Memory (VRAM): 128 MB
Shared Memory: 3995 MB
DirectX Features
DirectDraw Acceleration: Enabled
Direct3D Acceleration: Enabled
AGP Texture Acceleration: Enabled

Can it run in cpu (since no dedicated gpu bc laptop)

[hallow thistle]

Intel(R) UHD Graphics is an integrated GPU; not really recommended to be used with any AI program such as W-Okada.

[agile relic]

I can run it in the cpu

I think

[hallow thistle]

Yes, you can run W-Okada with only CPU but it will be really slow as hell; also not recommended.

[agile relic]

the cpu is kinda good I'd say for being a budget laptop

[hallow thistle]

[agile relic]

400$ laptop, scam?

[knotty moth]

you don't seem to have Nvidia/AMD gpu

[agile relic]

it's a laptop.

it will run on the cpu just fine don't worry

my cpu never failed me

[hallow thistle]

But it's not recommended, you know. There's an online option for this one.

[knotty moth]

https://tenor.com/view/fine-this-is-fine-gif-20102174

[hallow thistle]

[agile relic]

tell me more

do I run setup64 or setup exe for vac

[hallow thistle]

setup64.exe

[agile relic]

"I am an advanced user" ?

[knotty moth]

yeah 20+ sec delay seems fine for you

doesn't it?

[agile relic]

It was running 500ms delay with through mode it will be fine

in 192

[hallow thistle]

Also, don't expect to run the program with another game at the same time. You know right?

[agile relic]

it will, trust

also when I installed the mmvcserver it wouldnt work and I had to manually create some folders and empty jsons that were missing idk if this has to do with anything

[hallow thistle]

What the hell?

[agile relic]

my laptop is the best

[hallow thistle]

I don't trust you for this, seriously.

[agile relic]

I got the cable installed Imma try to figure the rest

I think I should reset my pc

restart

[hallow thistle]

If you really really wanna try W-Okada with CPU-only that bad huh, there's a better W-Okada version, which is in this link. https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#download-amd-intel-and-cpu-on-windows I just better hope this is for experiment and not anything else.

[agile relic]

nah I'll be fine, you'll see

[hallow thistle]

No.

[low shard]

It will run very bad, this is more intensive than a game, I can say it since I had previously a laptop with a similar CPU and same RAM

AI is intensive, it's not easy to run locally nor a site like chatgpt

[hallow thistle]

Any W-Okada variant, like NVIDIA and AMD/Intel (DirectML), that made for Windows can work with only CPU; the AMD/Intel one is somehow known to work with an integrated GPU like Intel UHD Graphics while NVIDIA one doesn't.

[agile relic]

the console keeps saying
Pipeline is not initialized.
Waiting generate pipeline...

[hallow thistle]

You can't be overconfident on this one.

[agile relic]

I still dont get volume in the virtual audio cable

[hallow thistle]

Deiteris' fork W-Okada should look like this. If you still use the version that has many files in the same folder, you'd better stop use that version.

If you change your mind and thinking to use the program with a game or Discord, I'd suggest to consider for an online option like Kaggle instead. https://docs.aihub.gg/realtime-voice-changer/cloud/deiteris-w-okada-fork-kaggle/

[knotty moth]

tell me which voice changer version you're currently using?

(first ask me for image perms to show the screenshot)

[agile relic]

I was using the latest nvidia 18a something

[agile relic]

I'm using this rn

"Waiting for the operation to complete..."

it loaded

[knotty moth]

if you seem to feel fine on it, then you wouldn't need our help anyway 😔

[agile relic]

I dont understand

[knotty moth]

same, I don't understand why you don't understand on it

[agile relic]

well the ping says over 11000ms and it keeps raising

[knotty moth]

so you've got the problem, eh

[agile relic]

I can change the settings

[knotty moth]

do you know what we were recommending you?

@agile relic

[agile relic]

yes I downloaded this and I'm trying it rn

the one with 15k ping

[hallow thistle]

v.1.5.3.18a and b2332 are not the same one.

[knotty moth]

15k ping on the version we gave you, but normally on your current .18a?

[agile relic]

I am trying this web version that has so much ping

"Download AMD, INTEL and CPU on Windows
The latest version as of December 7th 2024 is: dml-b2332 (click here to download)"

should I just stream

[hallow thistle]

Struggling huh, that's unfortunate, but here's the settings you can try:
Chunk: 300 ms
Extra: 2.7 s
GPU: CPU

F0: rmvpe_onnx

Now expect it where its audio being delayed more with higher audio quality at the same time.

[agile relic]

Is Line 1 supposed to never register anything it's just my microphone's volume bars that increase, but the Line 1 Virtual Audio Cable is not getting any signal

[hallow thistle]

Input: your microphone
Output: Line 1 (Virtual Audio Cable)
Monitor: your speakers/headphones

[agile relic]

Yeah and it doesn't work

"
Input: Microphone (TRM-10 Audio Device)
Output: Line 1 (Virtual Audio Cable)
Monitor: Headphones (Realtek(R) Audio)
"

just forget it, ig my pc is just dumb

[hallow thistle]

[agile relic]

I'll just stick to voiceweave or some shi

that one did work for me and it was just 200ms delay

[hallow thistle]

You should've give up expecting your laptop to be great the first time I told you to go for an online option lol.

[agile relic]

oh you meant this when you said online

[hallow thistle]

I gave you the other options, and you just ignored them. Now it's your choice.

[agile relic]

p2w

[knotty moth]

that means you don't need help from us, thinking "if it ain't broken don't fix it"

[knotty moth]

seems more like a cloud service tho

well whatever you'd experience is your own responsibility

[viral mason]

Never heard of voicewave, def worse than any rvc program if it doesn't use rvc

[agile relic]

it has the option to use dedicated gpu for quality or simply your cpu but some sound dogshit others are passable for the most part

"some" almost every single one

[agile relic]

sorry, I called it the input name, the program is EaseUS VoiceWave

[oak edge]

@viral mason gang you know anything about realtime voice changing apps?

[viral mason]

I know absolutely nothing about the code but I know how to use them, only Vonovox and wokada

[oak edge]

which one is good to use gang, imma try it

[viral mason]

If u got Nvidia I'd say use Vonovox as it's just better but only issue is it has only 8 model slots

That is the only flaw

[oak edge]

as in I can only put 8 epochs?

[viral mason]

No, you can only put 8 models in the voice changer at a time

Dr87 is working on that tho

Eventually

[ocean bluff]

I used ai to make me look like a chad in my selfie 🗿

[oak edge]

ohhhh it still works with the epochs I trained in rvc right

[viral mason]

Good for you 🧍

[viral mason]

Wdym epochs 😭

[oak edge]

idk how to use it

[viral mason]

It has nothing to do with training man it's you just put models in it that were already finished

[oak edge]

gang finished in where

im confused

[ocean bluff]

I also used AI to create a relistic guy pooping in a bucket

[oak edge]

like the models im trainin in applio rvc?

[ocean bluff]

And then he ate it

[viral mason]

Why 😭

[oak edge]

will this work?

[viral mason]

Like, any models, the ones in the voice models section, the one's on weights, any of them

[viral mason]

That'll work yes

[oak edge]

okie lemme tryyy

[viral mason]

Epochs aren't important on the voice changer because it's just using the model you have live

Instead of putting like a silly song in it to make it sing

[ocean bluff]

I made s relistic poop

It so stinky

[oak edge]

im confused lemme try and tell

[ocean bluff]

so shmelly and wet

[viral mason]

I wish I had time to show u how to use it but I got work soon

Btw u have to have vac lite downloaded

It should be in the guide

[oak edge]

it's alright I'll try a bit on my own and I'll ask you for doubts later on too

[oak edge]

OHHHH lemme download

oh it's just virtual audio cable

i already have it

[ocean bluff]

diareaa

[viral mason]

Ye

[ocean bluff]

I wonder why no one talks to me

Cause im so nonchalant and my farts are so stinky

[hallow thistle]

Because your statements made no sense, and this channel is not your typical general chat. Hope you understand.

[viral mason]

Shh don't interact with it

[hallow thistle]

Yes, I know about realtime voice changer.

[hallow thistle]

I'm aware he's just a kid who joking around.

[oak edge]

I just got vonovox but haven't downloaded files yet how many gbs will it be

[hallow thistle]

Vonovox won't gonna look like a typical W-Okada. Would you still go for this one? https://raw.githubusercontent.com/dr87/Vonovox/main/docs/images/ui1.png

Deiteris fork W-Okada has more friendly GUI, while Vonovox is more of professional.

[oak edge]

im confused i have nvidia gpu

[oak edge]

the terms block size and lookahead buffer is confusin

but whatever gives me best quality im ready to learn

[naive dune]

Has anyone ever used vibe voice?

[brittle bobcat]

will Kraggle ever be fixed? Does it have any relation to this server? its on aihub

[latent kettle]

I did

[naive dune]

Do u need to input/type in speaker in order for the ai to work for one person?

[latent kettle]

1.5B is bad. 7B (large) is good. For English. You can add multiple speakers and create podcasts

[latent kettle]

Do you mean by Audio output?

[naive dune]

Yes

Im basically trying to create or clone a character voice and trying to get realistic emotions

Idk if vibevoice is best for that

[latent kettle]

It's better than others but you can try chatterbox Too.

[viral mason]

what are you

[latent kettle]

Yes you can clone

[naive dune]

I see

Im using it for my manga and stuff

[latent kettle]

Maybe chatterbox performance is better. Wait I'll send you demos

[naive dune]

Okay!

[latent kettle]

https://resemble-ai.github.io/chatterbox_demopage/

[naive dune]

thx

[low shard]

I'm guessing it's a typo of Kaggle,
Kaggle is just a Cloud (remote good pc) service, we can't know which notebook you're talking about since anyone can code one and this is a general ai server
Please elaborate more :)

[brittle bobcat]

it's the one on aihub that links to it. should i link it?

[latent kettle]

What do you think? How it is

[low shard]

If you're talking about https://docs.aihub.gg, this is our documentation, it contains various guides and programs, we can't know which specific kaggle notebook you're talking about, since we help with all programs
It would be better if you elaborate more your PC Specs, the error and the link to the kaggle notebook

[naive dune]

Give me a sec Im gonna try to install it

[latent kettle]

Well you will get an error if the visual studio is not installed. Building wheel error

Also hit a try to huggingface space here https://huggingface.co/spaces/ResembleAI/Chatterbox

[naive dune]

Im planning on using it in ComfyUI

Okay

[latent kettle]

Hmm, is it natively supposed in comfyui ?

[naive dune]

looks like it, ill update u on what happends

[latent kettle]

No it requires custom nodes. I guess

[merry forge]

i js tried training my first model on rvc but i didnt get the pth, how do i get it?

[latent kettle]

/Logs

Check logs folder 📂

[merry forge]

theres only an index in the logs folder of my model

[latent kettle]

Are you using mainline or applio

[merry forge]

i usually use applio but im using this one now https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI?tab=readme-ov-file

[latent kettle]

Check in assets\weights

@merry forge

[merry forge]

theres no pth with the name of my model

[latent kettle]

Have you checked "save every weight " In GUI ?

[worn jungle]

How to use voice model?

[merry forge]

yes

if u mean the save small model

[lusty whale]

does anybody know why your not able to hear yourself or anything come out from wokada

or is it just me

[brittle wing]

o. o

hey

[knotty moth]

I don't even think if rvc comfyui is well maintained as the standalone Applio

[timber heart]

For the Kaggle Applio notebook, what's the format for the dataset file path?

[sonic bear]

im new to this ai stuff i made a song but i want it to sound real as possible like ynw melly voice can i send it to someone to help

[timber heart]

Ah man, now I feel like a fool; I had leftover datasets from a year ago, I just forgot about how to do that part lol.

Thanks man.

[viral mason]

you're good, some things never get explained in guides or are easy to skip over

[hexed osprey]

Yes, a guide would be great, I am on windows 11.

[hallow thistle]

!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 lega, 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.

[hallow thistle]

RVC or W-Okada realtime voice changer?

[low shard]

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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

[low shard]

You can choose either wokada deiteris fork or vonovox

Read each pros&cons

[worn jungle]

RVC

[simple ore]

or you can use dataset creator from UI

[latent kettle]

Did you installed it ?

[timber heart]

Just realize that now, oops.

Thanks for that as well.

[elfin nebula]

So I already trained my voice model to have a neutral voice, but how do you give it a wide range of voices that emulate emotion?

[mortal shell]

best rvc model

wth ai not working

[agile lynx]

its with ai helping stuff

not the ai heliping

the people is gonna help you not I

ai*

[viral mason]

can we give homeless guy the award for most useless help message in this chat

[twilit jasper]

Hey! I'd like to upscale and interpolate a video from compressed and grained 1080p 30fps to clean 8K 60/120fps. The video is a gas giant in space, so with a lot of black and a lot of big compression blocks.
I tried multiple models (OpenProteus, Real-ESRGAN and another I can't find in my mind) but every models has an issue.
Open Proteus has a blurred result for great colors, the keyframes of the video are pretty visible (even more when slowing it down) and the compression was not perfectly concealed.
The other model I can't remember has good details but a bit too much contrast, the real-looking video almost became like an anime, and the keyframes was no longer so much visible, but the compression was looking blocky and weirdly colored.
And Real-ESRGAN is very slow (It's not a problem for me if the result is good, I can keep my computer running for a few days if needed) and has wayy to much contrast, it's anime-looking (But I feel like it's typical Real-ESRGAN, almost every photos I upscale with it has too much contrast)

Can you recommend me a good model for upscaling compressed videos?

(I don't need interpolation in first time, I can do it after upscaling and most interpolation models are pretty good)

[compact sorrel]

how to train models/voices?

[naive dune]

Sorry work got in the way, I’ll update u on what happens

[sly valve]

I’ve made sentences where I say gibberish, and Speech To Text translates it to what it thinks I’m saying. What channel should I share it on?

[burnt shell]

hey guys so i installed a voice and when i talk it just like freezes mid talks

[midnight fern]

Does anyone have a working link for okadas voice changer? Bc my pc isn’t too good for the program

[viral mason]

what gpu do you use, Nvidia, Intel, AMD?

[broken urchin]

hey i was wondering

what's the difference between one big dataset file for RVC voice training

like one big one with 90 minutes

or the same one but its just split in 50+ different audio files

what's the difference?

does it produce better or worse results or

[viral mason]

I'm not entirely sure, I use just one thing of audio but I think if you select the post option in normalization it auto cuts the dataset into smaller files

[midnight fern]

Im on a shitty mac even the kaggle link had delay for me

Is there any links?

[viral mason]

If the one on kaggle didn't work I think you're out of luck unless you can afford to upgrade 🙁

[broken urchin]

so it doesn't matter much right

even if the audio isn't split

[dim peak]

how to i change epochs on rvc

[viral mason]

Again I'm not entirely sure, I'd ask other ppl who train models to get more opinions on it

[frail bolt]

Hi, i am searching for an AI to sort my mail, for example Gmail.

[simple ore]

no

[compact trellis]

guys when i downloaded the main voice changer and i try open “http_start.bat” it doesn’t work for me what do i do?

[knotty moth]

how to change my birth date

[knotty moth]

it doesn't matter if you make sure to use preprocess settings like this

[willow lintel]

Anyone know what app would be suitable for a voicechanger in mac i5? Ive been trying so hard but no results. It would be great if someone also gave me a tutor

[latent kettle]

Okay not a problem

[naive dune]

https://github.com/resemble-ai/chatterbox Im trying to install it using cmd but I get an error

[latent kettle]

What error?

[naive dune]

let me send it

[latent kettle]

Send the full error

If it says unable to build whell, install visual studio 14+

Than you have to download Microsoft visual studio and install some frameworks

[naive dune]

ok

[latent kettle]

Which python version are you using

[naive dune]

25.1.1 im upgrading it right now

[latent kettle]

Python --version

Try 3.10 or 3.11

[naive dune]

okay

Its the same

[knotty moth]

try using python 3.11 instead of 3.13 you're having

[knotty moth]

and please show the error message, I bet it must be different

[naive dune]

Im trying to get that done right now and set it as path, give me some time sorry

YESYESYESYSEYESYYESYES

ITS WORKING

ITS INSTALLING

I set it to path

[naive dune]

Okay I have it installed I just need to open it

[marsh raven]

@knotty moth are you able to help figure out VAC its not being detected even in discord

[broken urchin]

are those the default settings?

[latent kettle]

Hey I got error on building wheel

[naive dune]

Update: after 10 trillion times on fixing it, it was dependencies that did not come with the package

So I used chat gpt to help me 💀

And it worked eventually

Also for some reason it’s using my cpu now which is weird so I gotta change it to use my gpu

[simple ore]

chatterbox can run with 3.11 or 3.12, you need to modify pyproject.toml to fix numpy dependency to 1.26.4 and remove pkuseg

and for gpu use you need to run torch install with cuda

[willow lintel]

Anyone know what app would be suitable for a voicechanger in mac i5? Ive been trying so hard but no results. It would be great if someone also gave me a tutor

[short torrent]

I set deteris w-okada fork for cuda and everything and the voice changer seems to work fine in all of the tests but whenever I talk over discord it shows that there's no sound? I have a GTX 1070 Ti I know it's not the best but when I test my mic within discord settings it sounds perfectly fine and everything runs smoothly, it's only when I try talking in a voice call or within a game where there's zero sound being heard or indicated
There were a few threads about this issue but none had an answer so I was wondering if anyone experienced knows why this is happening?

[weary socket]

anyone knows if the training script from applio is different from the original rvcwebui?

[jovial spire]

question lol 1-Does anyone here use voice.ai and 2- have people been getting the "No network connection, voice initialization failed. Please try again #3769" error? I can't find any way to see if their servers are down :/

[weary socket]

I wonder if using applio instead could improve my model

Also I wonder what happens if I train directly using multiple voice sources, in contrast to using model blender(average the weights) of multiple models

[simple ore]

one one voice per voice model

[viral mason]

Just saying voice.ai is worse than both w-okada and Vonovox, and it's pay to win in voice.ai both I mentioned are free

If u use an Nvidia GPU u should try out both but if u have some other one like AMD or Intel you could use the first link/guide

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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

[jovial spire]

I’ve had success with it so far haha I have like 200k credits for some reason lol I just wanna be a cringe ass eboy online so people stop going “mommy? Sorry. Mommy? Sorry” 😫

[latent kettle]

Hey I got error with pkuseg. How to fix it. There is no wheel file already built for it. And I don't want to install that visual studio thing

[viral mason]

[simple ore]

remove it from pyproject.toml

[latent kettle]

And what to add instead ?

[simple ore]

nothing, if you dont need chinese it should be fine

[latent kettle]

I see. Thank you. Is there anything I have to do with numpy ?

[simple ore]

with python 3.11

you can install as is, you just need to clone the repo, update pip setuptools wheel packages

then install numpy standalone, then you can use pip install .

for python 3.12 you need to modify pyproject.toml and change numpy version to ==1.26.4 instead of the existing stupid requirement

then it installs just fine

[latent kettle]

I see, thank you again

[pine smelt]

Dunno if anybody has experience with this, but is there a reason why in the tg-dev fork I cannot select GPU under Processing Unit? My only option is CPU

[simple ore]

What is your GPU? screenshot task manager/performance screen with the gpu selected

[pine smelt]

In deiteris I can and did select my 4090 no problem

but here I seem unable to do so

It's just not an option in the select field

[simple ore]

you got the right fork?

[pine smelt]

Uh, I think so but I might have not?

[simple ore]

what did you download?

[pine smelt]

Lemme check, perhaps I picked the wrong one from the release page?

Oh damn that's it, sorry for being such a noob

[simple ore]

Download NVIDIA on Windows

Download all of the voice-changer-windows-amd64-cuda files. This will include multiple files, likely ending in .zip.001, .zip.002, etc.

[pine smelt]

your name nickname would be more appropriate for me in this case damn

sorry for wasting your time!

[quiet hatch]

Agentic AI Framework (agenticaiframework) is a next-generation Python SDK for building agentic applications with advanced orchestration, monitoring, multimodal capabilities, and enterprise-grade scalability.
It offers a modular, extensible architecture for creating intelligent agents that can interact, reason, and execute tasks across multiple domains — from simple automation to complex multi-agent ecosystems.

https://github.com/isathish/agenticaiframework

[short torrent]

anyone?

everything is up to date & i'm using the right virt cable

[simple ore]

may want to use push to talk in discord

[short torrent]

i'll try it

[knotty moth]

so have you got the correct installation?

[weary socket]

So I have 3 voice source for 3 different languages, and I want to make one model out of it. I don't think blending would be a good idea

[simple ore]

you not gonna get what you expect to get

[weary socket]

Is there some out of the box mixture of export for this purpose?

[simple ore]

you can train 3 separate models then merge/blend them with each other

[pine smelt]

I didn't. I downloaded the right one and the issue got fixed

[jovial spire]

Hey is there a ticket system? I have a mod question

[hidden hinge]

few questions im quite new to whole real time voice changer thingy... so how much quality is lost if you speak in language model wasn't trained on, what is good few voices i could get to play around with them, and what are the best settings for mmcv ?

[viral mason]

what gpu do u have, also what voice changer are u using, there's two Vonovox and Deiteris Okada

if u got the voice changer from a yt video it's probably over a year old btw, and for ur question pretty sure if using a different language on a model trained on something specific doesn't change quality

[low shard]

Dm AI HUB Modmail :)

[hidden hinge]

I do have 3060 mobile 6 gig of vram and I think I got this one ?https://huggingface.co/wok000/vcclient000/blob/main/MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a.zip

[viral mason]

3060 mobile? is that Nvidia, AMD, or Intel

[hidden hinge]

It's Nvidia just laptop version

[viral mason]

pretty sure Vonovox will work!

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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]

it's the best realtime voice changer out there (rvc doesn't stand for that) but if that doesn't work just download deiteris

[hidden hinge]

When I hop on laptop imma test it out any voices/settings you recommend ?

[viral mason]

for vonovox I'd use these
0.3 block size, 2.0 second extra, 0.15 crossfade

and if u need to switch to deiteris use these

[viral mason]

for voices, depends on what u want but u could test my Venom model out, just made ithttps://discord.com/channels/1159260121998827560/1416114563845324930

[hidden hinge]

how much index file influence voice ? and if i can ask what will change if you swap audio backend, f0 method, RNN noise and upscaler (im guessing that last one make it slower but better)

[viral mason]

it's best to not touch the index, just import it in but don't use it as vonovox has had issues with index causing exreme gpu/cpu usage in the past

[hidden hinge]

also dumb question how do you eject file?

[viral mason]

keep smart sine on, not sure what you mean by this tho "what will change if you swap audio backend" and keep f0 at rmvpe

[nova willow]

does anyone know good ai voice cloning stuff that I could use for voice modding that is low gpu usage

[viral mason]

oh I don't use that feature so I got no idea

[nova willow]

I am bored

[hidden hinge]

in audio device setting you have "audio backend" and im curious if it changes much or its just backup system

[viral mason]

wdym voice modding, are yu looking for a voice changer or to make ai voices of existing characters/people

[viral mason]

wasapi is recommended

[nova willow]

voice changer

I want to use it for horror games

since I can't afford a voice actor

[viral mason]

make sure your voice settings are
input: regular mic (headset, headphones ect)
output: vac lite
@hidden hinge

[viral mason]

what gpu do u have, Nvidia, amd, intel

[hidden hinge]

and what is F0 method or i shouldn't really bother ?

[nova willow]

I got a 3050

[nova willow]

from nvidia

as I said I am poor

[viral mason]

that should work fine with either of these, there's a guide for both but f u want good quality go with Vonovox, there's a guide here, it's what the other person is using too

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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

[hidden hinge]

yee it doesn't use much of my vram so you should be able to run it on 3050

[nova willow]

what is the one that has the least amount of artifacting

[azure patio]

guys im lost what rvc is the best to use on web?

[viral mason]

idk what it really does but use rmvpe, the creator of this vc said that swift is good but it's still kinda new

[nova willow]

I don't like when I hear ai voice artifacting

sorry if I am being harsh

[viral mason]

ur ok

use Vonovox, it's got a lot of settings to prevent that kinda stuff

[azure patio]

[viral mason]

what are u looking for, a voice changer or to make model

rvc doesn't mean realtime voice changer btw lol

[azure patio]

i have my models so voice changer

[azure patio]

yep ik

[lethal iron]

which tutorial do you guys use for comfyui zluda windows 10

[true iron]

hi, can you guys help me? I try to find good RVC for voice change

[viral mason]

Depends on what gpu I have, Nvidia, AMD, intel

[nova willow]

also since I am on the topic of game development and I can't hire a 2d artist/don't wanna dox people, what should I use to make realistic images (the game I am making is like welcome to the game)

[viral mason]

What gpu do u have

Nvidia, amd, intel

[viral mason]

I don't personally know but if you are making a game I'd avoid using AI art in it as could get a lot of hate, u could use a drawing software and teach yourself on a free app like medibang pro

[nova willow]

only reason why I am asking is to not use real humans and not hurt people, welcome to the game is based on the dark web

I like the game so I wanted to make my own version of it on unreal

not drawing anyone

[viral mason]

That's interesting

[nova willow]

I have reasons to do stuff I wanna do

since I am a gamedev

and I believe that fiction is fiction

and shouldn't go into the real world

[viral mason]

Y'know you could put in the lore that an ai is making the art so you have an excuse for ai art

[nova willow]

I mean real life looking human beings

[viral mason]

Just a suggestion of course

[viral mason]

O

3D humans orr

[nova willow]

and I want to have the same type of person so I don't want it to have overlap and look like a completely different person

[sonic garden]

-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

[nova willow]

only pictures

its web browser based

[viral mason]

Ah ok, I'm not sure tbh what you could use

I'd ask around

[nova willow]

ty

I just don't like using real humans in works of fiction so that is my reasoning

espesically of this topic

[viral mason]

I don't really understand it but no judgement from me 🤷‍♀️

[nova willow]

Ill send you a vid of the game

so you can understand why I DON'T want to include real humans in this

also the creator themselves use pictures of real women and a lot of people didn't like that

and I agree with them

https://www.youtube.com/watch?v=1MxkK583uWQ&t=4s

[hidden hinge]

well it really depend on what gpu do you have and how many really similar photos of a character you need

[nova willow]

my gpu is 3050 if there is a cloud company that I could use I could pay for it and enough to make fake dark web social media (basically how you talk to other hackers with stolen info)

[hidden hinge]

hmm then you have few options you can install stable diffusion model that is good at realistic or flux model (second probably better for realistic) or use some websites that give you free daily use of img generation and use something like flux kontext or qwen image edit to modify already generated by you photos to get new variants

[jaunty walrus]

Anyone got this working on NixOS?

Even with FHS I couldn't get Deiteris' fork working:

ImportError: /home/xnefas/Documents/AI/MMVCServerSIO/_internal/libstdc++.so.6: version `GLIBCXX_3.4.29' not found (required by /home/xnefas/Documents/AI/MMVCServerSIO/_internal/onnxruntime/capi/onnxruntime_pybind11_state.so)

[nova willow]

bro

I spent 5 bucks

and it has NOTHING

[simple ore]

sudo apt-get update

sudo apt-get install gcc-4.9

sudo apt-get upgrade libstdc++6

sudo apt-get dist-upgrade

[upbeat thunder]

hello

#✨│ai-help HELP

[balmy trench]

Hello, everyone. Where can I try seedream v4 edit? I couldn't find it on Hugging Face.

[upbeat thunder]

bruh how yall doin this

[weary socket]

I got following message after running the training script: "Not enough data present in the training set. Perhaps you forgot to slice the audio files in preprocess?" I do have sliced the audio, and roughly 400 audio files in the sliced directory. what could I miss?

[hallow thistle]

If you left before a helper came in, it's your loss.

[weary socket]

oh, my f0 directory is empty...

I thought the extract.py would do the feature extraction

[hallow thistle]

Why did you pay for this?

[rare flint]

where's my RMVPE

[hallow thistle]

"rmvpe" is only available in NVIDIA variant of W-Okada, and it doesn't exist in DirectML of it. Judging by your screenshot, it seems like you're using the "original" version of W-Okada DirectML.

[knotty moth]

I would recommend getting better version here (depending on your gpu)

https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#windows

[nova willow]

it worked nvm

[dry coyote]

yeah

i went through a few forums

and the reason why its outputting mumbling is because i was using a spin model which deiteris's fork of wokada doesnt support

neither does the base version

but apparently i can implement a SPIN embedder

https://github.com/deiteris/voice-changer/pull/213

but idk how to do that

[hallow thistle]

I can implement a SPIN embedder
but I don't know how to do that
Really?

[dry coyote]

yeah like 70% sure

did u check the pull req

[simple ore]

there's a new fork with that already

#📰│dev-updates message

[dry coyote]

is it a fork of deiteris's fork?

[simple ore]

duh

yes

[hallow thistle]

For another fork W-Okada, you might wanna look at this one. https://docs.aihub.gg/realtime-voice-changer/local/tg-develops-w-okada-fork/

[dry coyote]

nvm i dont think it is

[simple ore]

there's spin-v2 now

[dry coyote]

its just a fork of wokada

[simple ore]

"Wokada Tg-Develop's fork since it's a fork of the Deiteris' Fork, containing the performance improvements, has an improved Web User Interface, supports Spin Embedder Models, and has Audio Effects."

[dry coyote]

wheres that

[hallow thistle]

You can't judge it like that.

[simple ore]

read the link above

[hallow thistle]

[dry coyote]

wheres that

ah i see it now

[hallow thistle]

You didn't see my link?

[dry coyote]

i dont read i just download whatever

no hablas engles

[hallow thistle]

That's an issue.

[dry coyote]

its whateverrr

[hallow thistle]

Still.

[dry coyote]

damn this fork is big download

3gb

ts gonna take like 20 minutes

[hallow thistle]

Almost every AI program is as big as that in GB. This specific fork W-Okada, as you'd imagine, would pack with many features in there. Deiteris fork, however, certain packages and features were removed or missing when those didn't need for most users.

If your current internet is slow, it's gonna hard to get it done lol.

[viral mason]

fr

[dry coyote]

is 10bytes per second good

😼

[hallow thistle]

Seriously? This internet speed could download smaller text files or certain files in the past, but not other heavier file of today.

[dry coyote]

oh btw is it worth it to get a second gpu for just wokada

and if so whats the best one to use for primarily for it

in terms of value

[hallow thistle]

My internet is around 120Mbps, while not the fastest, it can download files in gigabytes in not hours but rather minutes.

[knotty moth]

RTX 2060, 3060, or 4060 depending on your budget

4060 would benefit AV1 streaming

[dry coyote]

how long is latency for 2060

[dry coyote]

wdym

[knotty moth]

ask google/chatgpt

[dry coyote]

😭

[dry coyote]

wait so its just better for transmitting audio packets over the internet??

[hallow thistle]

NVIDIA GeForce RTX 2060 is the minimum for most tasks, RTX 3060 is in between, and RTX 4060 is faster, depending on your budget. RTX 50 is the newest, but requires a specific fork W-Okada in order to work, which the Tg Develops one is not yet known to have the version for this GPU.

[rare flint]

uhh? i tried downloading the new one

[hallow thistle]

AV1 is a video codec. You might not need to know now, but only needed if you want to encode a video with higher quality possible some day.

[hallow thistle]

Did you download a correct fork W-Okada?

[rare flint]

yes, i think it's because forgot to download all 3

[hallow thistle]

Your GPU is NVIDIA GeForce RTX 50xx?

[knotty moth]

get either of these for RTX 50-series:

• https://docs.aihub.gg/realtime-voice-changer/local/tg-develops-w-okada-fork/
• https://docs.aihub.gg/realtime-voice-changer/local/vonovox/

[rare flint]

yeah, it's a 5080

hold on i think it's working

[knotty moth]

and you used the og directml version that would perform like RX 580 or just throw some error messages

[rare flint]

it's working now, troubleshooting for myfriend (and yes it was the DirectML issue)

[harsh oriole]

Good day!
I need help about the changer it does not apply to discord...

[hallow thistle]

!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 lega, 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.

[harsh oriole]

I see thank you.🙂‍↕️

[hallow thistle]

What is your PC GPU? Did you follow any tutorial before this?

[jaunty walrus]

Sorry but it's not Ubuntu, NixOS is one of these weird exotic distros that does everything differently

[hidden hinge]

never used nixos but wouldn't you have to add all dependencies to nix os config and regenerate system so it will grab all packages ?

[keen badger]

Unsupported post request. Object with ID 'xxxx' does not exist, cannot be loaded due to missing permissions, or does not support this operation. Please read the Graph API documentation at

GUYS HOW CAN I FİX THİS, WHATSAPP CHATBOT VİA FACEBOOK DEVELOPERS

[limber tangle]

-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

[boreal mango]

hi! how would i go on about training a voice model?

i wanna use it for narration

literally have no time to read

[low shard]

RVC is STS (Speech-To-Speech), not TTS (Text-To-Speech) natively, unless you use another TTS to make an audio output to use as an input in rvc (which is what for example the Applio RVC fork does automatically with Edge TTS, which is multilingual and good but emotionless)

[boreal mango]

hmm so i'd need a tts to narrate and put the narration into the rvc?

[oak edge]

yooo can someone help me out with formant shifting? (about how it works and how to use) for context i have male voice and trying to apply that voice for female portion of the song

[low shard]

So I got an AMD Radeon 860M gpu, should I run a voice model on cloud?
yeah, your gpu isn't the best for local AI

i'm guessing you want a realtime voice changer for calls/games (not pre-recorded audios) and are atleast on windows 10/11

-realtime

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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

[low shard]

you can use wokada deiteris fork or tg develop modified version

[low shard]

i mean, do you need specific voices or do u just want a general TTS?

u could just use a TTS alone

[boreal mango]

those sound awful

[boreal mango]

yup

[copper ferry]

Hi, I need help with RVC.
I believe my RVC is not functioning as intended. It does not change my voice at all. the delay is also extremely inaccurate. It's my own voice no matter what I change. It also does not let me hear myself unless I use pass through option.
The GUI of it is also very different with what I've seen from others.

v 1.5.3.16a onnxdirectML-cuda

I also have an AMD GPU, so if I should download a newer/different version of it or do anything to get it to work please tell me.

[viral mason]

-rt

[patent trellis]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• 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 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]

use the first link

download the amd version

[humble vortex]

What about actual voice models? Like voices. What are some of the best/most popular and most importantly trusted models here? I want something standard, grounded and realistic above all. Not female since I don’t wanna accidentally catfish anyone.

[low shard]

There are thousands of models that you can search in #1175430844685484042 and https://weights.com/models, but there isn't like a "common guy voice", the models are trained off a voice

[sour sun]

Hi, is the Spin pretrain supposed to sound like garbled mumbling, even when I'm using the new LegacyCore spin pretrain and have selected the Spin feature extraction?

[simple ore]

you need to prepare your dataset with spin embedder and infer with spin embedder

[analog obsidian]

i'd not use that pretrain because it got trained when the discriminators were broken

[simple ore]

also there are two of them now (spin and spin-v2)

[sour sun]

Hmm lemme check

I renamed it a bit for ease of use but I downloaded them pretty recently, unless there's another page I didn't see 🤔