✨│ai-help

[hardy yew]

Hi! How do I get the model maker role? The "Submit Model" button in #📤│model-maker-role just says "This interaction failed", and the guide on https://rentry.org/fdg_guide_new looks outdated

[feral sierra]

I'm no expert but from what I've experienced original wokada is pretty outdated, I'd recommend using tg develops fork for local realtime conversion

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

[tawny viper]

thank you

[hallow thistle]

Check if you have these settings enabled, because if you have these disabled, the bot can't direct message to you.

[tawny viper]

how can i download i cant find sorry

[hardy yew]

sadly everything already active in my case

(and here's the result after pressing the button)

[hallow thistle]

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

[sweet ridge]

#1474755224538189915

can someone help

[hallow thistle]

Well, it can be the bot itself is offline. Maybe try again next day.

[hardy yew]

Will do, thanks for help!

[feral sierra]

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

[tawny viper]

i find but thats what i already use

[hallow thistle]

Tg Develop and Deiteris, while they both made their own W-Okada versions respectively, have different interfaces, despite sharing the similar naming schemes, just check their zip file sizes.

[tawny viper]

oh mb oky

[tawny viper]

i have ping again

help

[hallow thistle]

An AMD Radeon GPU from 2014 can work with W-Okada voice changer? While it might work, it would sure struggles even hard. The minimum AMD GPU to run with the voice changer would better be those "Radeon RX" GPUs, especially "AMD Radeon RX 580" or ones that have at least 6 - 8GB VRAM.

[tawny viper]

oh oky mb

[hallow thistle]

The crazy story, those NVIDIA GeForce GT/GTX GPUs (such as GT 730) prior to GTX 9xx series often fail on voice changer regardless of any version, but when I see a 2014 AMD Radeon GPU working with W-Okada DirectML, it indicates that even older AMD GPUs might have their potentials to work with modern AI programs, though not always recommended as an actual practice anyway.

[edgy urchin]

mine doesnt even work

it gives me errors when i try to start the realtime

[vague timber]

Hi, i'm having troubles making the VC work, I've installed everything but when I select the VB cable on discord, it doesn't emit sound, it's like it doesn't register my mic. Am I doing something wrong?

https://i.gyazo.com/86c607398ab06e31b39bf8bea8a4e2bf.png

https://i.gyazo.com/28bebc4a2598b8be1484d4b25459f7a5.png

I have this settings but I can't seem to make it work.

Also, is this the best version for real time voice changing? Like for roleplay and stuff.

[hardy yew]

does it work when you set output to your speakers? Can you hear the voice changer then?

also, what you're using is very old

-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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[hardy yew]

see vonovox or tg-develop

[vague timber]

nope, I can't hear it neither like that

[vague timber]

i'll check it out, thanks!

[edgy urchin]

can someone help me install applio voice converter or however u call it

ive been running it on kaggle always but ts sum times glitches and its annoying

i wanna run it localy

[hardy yew]

oof

[edgy urchin]

im not a female also

[hardy yew]

see the message from bot above, you have list of local reatime RVC apps as well as links for guides there

[exotic glen]

ai

[edgy urchin]

no point on deleting the mssg

😭

[exotic glen]

why you ssd it anyway

lmaoaoao

[edgy urchin]

k

[edgy urchin]

i did before saying im not female cuz i knew u was gon delete it

[hallow thistle]

This goofy message might end up in #🥵│hall-of-shame for sure.

[loud summit]

One message removed from a suspended account.

[thorn zinc]

is it really bad that my perf is 2600 to 1740 ms

how do i lower it

[lilac hollow]

How do I stop hearing myself in RVC

it's really distracting

[hardy yew]

disable monitoring

[lilac hollow]

It's set to none

[hardy yew]

did you set your output to speakers then?

[lilac hollow]

Input is set to my mic
Output is set to VAC

[hardy yew]

did you enable "listen to this device" in windows audio settings in VAC properties?

[lilac hollow]

not enabled

[hardy yew]

check in audio mixer what makes the sound then, something for sure does

[lilac hollow]

yeah nothing's working
I'm just going to try and reinstall everything

[rare siren]

https://huggingface.co/wok000/vcclient000/tree/main which one of these do i download for intergrated grapichs intel

[lilac hollow]

Ok so I managed to fix it
Even after reinstalling everything I had the same issue
I just switched to Chrome instead of Firefox and now I can't hear myself just like I wanted
Thanks for the help

[hardy yew]

you can only run it on CPU then

[rare siren]

there is no gpu at all?

[quasi condor]

nah

u gotta have an actual gpu to use okada

or use cloud

[rare siren]

nah but the link i send

[rare siren]

ths one @quasi condor does it have gpu one and which one is for intergrated graphics intel XE

[quasi condor]

id suggest u use cloud

using the voice changer on a intergrated gpu is a bad idea

[rare siren]

how can i do that is there a yt tut ?

[quasi condor]

@hallow thistle

[viral mason]

Weights has disabled uploading because they're shutting down in march

[finite saffron]

hi

[hushed mica]

hola

guys whats the difference between testing and validation data

im told in my homework that i need to cut up my dataset to training, testing and validation

i understand the training data, although i cant really tell the difference between the other 2

[gentle skiff]

No module named 'sox'

[hallow thistle]

Hi. Is there something you'd like to get help with?

[finite saffron]

you ai??

[hallow thistle]

Why would I?

[hallow thistle]

Remember that the same "user" refused to go for newer W-Okada version and still sticks with older W-Okada in the thread https://discord.com/channels/1159260121998827560/1471239901612740743 . I doubt if he ever changes his mind and opens to new things.

[wind cosmos]

i keep getting FileNotFoundError: [WinError 2] The system cannot find the file specified: 'model_dir\4\' error

highkey the file exists

how do i fix

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[wind cosmos]

NVIDIA RTX 4070 8GB VRAM
Windows 11
I downloaded a model from here, and I'm now attempting to upload it to the voice changer client app, but keep getting an error not letting me upload / therefore use.
I was using this guide https://www.youtube.com/watch?v=SxdnGxicJOg&t=26s
In the tutorial I messed up at 4:50 seconds.

I would upload a photo, but I cannot here.

The primary error when uploading is FileNotFoundError: [WinError 2] The system cannot find the file specified: 'model_dir\4\'

I restarted my pc and changed file locations, I feel a little stupid doing this currently

I have MMVCServerSIO_win_onnxgpu-cuda_v.1.5.3.18a.zip

[hallow thistle]

Tutorial videos from YouTube likely telling you to use the older W-Okada versions, which not always recommended. Try Tg Develop's W-Okada voice changer fork instead. Download https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-cuda.zip.001 and https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-cuda.zip.002, both split zip files have to be in the same folder, use WinRAR or 7-Zip to open and extract .zip.001 one.

[wind cosmos]

So I download both, throw both in the same folder and unzip contents in the same folder, and then redo the normal process?

[hallow thistle]

Similar to the older W-Okada, extract "MMVCServerSIO" to somewhere like "D:\MMVCServerSIO", do not extract anything on your desktop folder.

No, just open the .001 one.

[tacit moth]

whenever i run startup batch it opens command prompt for like a few seconds and then closes and doesnt open anything

[wind cosmos]

dont run as admin

[wind cosmos]

Alr bet i got the window up now, but i still cant upload and i dont have a debug console bc of it

[tacit moth]

still nothing opened after

[wind cosmos]

Restart pc

highkey

just keep retrying

[tacit moth]

i already restarted my PC

[wind cosmos]

make sure antivirus doesnt flag, make sure all files are in one folder, dont run as admin

[tacit moth]

how do i make sure its not flagged anything?

[hallow thistle]

What is your PC GPU? And did you follow any tutorial or guide before? I know this is always have to be the voice changer, though I ask for confirm.

[tacit moth]

i have an nvidia GPU and i did follow a guide

[hallow thistle]

Which NVIDIA GeForce RTX?

[tacit moth]

Nvidia Geforce rtx 4070 ti super

[wind cosmos]

This is ragebait, im in a new ui called RVC and Im getting the error

Error
Failed to upload model. The model file was not properly saved.

This happens when trying to upload in the web versin

It still throws the FileNotFoundError: [WinError 2] The system cannot find the file specified: 'model_dir\0\'

[hallow thistle]

My guy, it ain't that deep. When I extract "MMVCServerSIO" to drive E, I launch the program and it still works. What do you mean it won't work?

[wind cosmos]

Give me image perms

[hallow thistle]

So which guide? There are Tg Develop W-Okada, Deiteris W-Okada and Vonovox.

[hallow thistle]

I'm not a moderator.

[tacit moth]

can i send u the link to the video i was using as a guide?

hold up i can just send a screenshot

[hallow thistle]

Well, it doesn't sound like you were following any guide at https://docs.aihub.gg/, where the website has W-Okada forks and Vonovox there.

[wind cosmos]

highkey hop on quick call in weights in this server

[tacit moth]

and how do i tell what one to install bc following the vonovox one doesnt specify

[wind cosmos]

2026-02-21 19:23:30,784 INFO [main] Python: 3.12.10 (tags/v3.12.10:0cc8128, Apr 8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
2026-02-21 19:23:30,785 INFO [main] Voice changer version: b2397 NVIDIA-CUDA
2026-02-21 19:23:30,786 INFO [main] Server settings: model_dir='model_dir' sound_dir='sound_dir' content_vec_500_onnx_on=True host='127.0.0.1' port=18888 ssl_enabled=False ssl_certfile=None ssl_keyfile=None allowed_origins=[] edition='NVIDIA-CUDA'
2026-02-21 19:23:30,786 INFO [main] Checking for mandatory models...
2026-02-21 19:23:30,897 INFO [ModelManager] Found 3 missing mandatory models. Starting download...
2026-02-21 19:23:30,897 INFO [ModelManager] Downloading mandatory model: RMVPE
2026-02-21 19:23:30,897 ERROR [ModelManager] Failed to download mandatory model rmvpe: [WinError 2] The system cannot find the file specified: 'pretrain'
2026-02-21 19:23:30,897 INFO [ModelManager] Downloading mandatory model: RMVPE (Onnx)
2026-02-21 19:23:30,897 ERROR [ModelManager] Failed to download mandatory model rmvpe_onnx: [WinError 2] The system cannot find the file specified: 'pretrain'
2026-02-21 19:23:30,897 INFO [ModelManager] Downloading mandatory model: ContentVec / Hubert
2026-02-21 19:23:30,898 ERROR [ModelManager] Failed to download mandatory model hubert_base: [WinError 2] The system cannot find the file specified: 'pretrain'

It cant find or make any folder file names. This is when I am extracting the files you gave me namari

[hallow thistle]

https://github.com/dr87/Vonovox/archive/refs/tags/v1.6.9.zip

[hallow thistle]

Try talk anything until your username turns blue, and you would be able to send an image here.

[wind cosmos]

im going to spam then delete it, if that works - unless theres spam prevention

ay

ay

ay

conversation

ay

[true grove]

i got timed out, who woulda thought

[tacit moth]

i open the startup batch and then use the launcher.py to launch?

[true grove]

I would use setup, then start

Although, I still get errors. Which could just be errors of my own, so go ahead and try

[tacit moth]

wait use start batch after i used setup?

[true grove]

wouldnt that make sense? set it up, then run it?

[hallow thistle]

Who would have thought it was a good idea to spam? I didn't tell you to spam either, I told you to genuinely talk about things and not a forced strategy.

[wind cosmos]

😭

namari, can we go into a vc, you can mute, and I show you what im doing

[hallow thistle]

To launch Vonovox the first time run, first run setup.bat and then start.bat.

[wind cosmos]

^

[tardy jungle]

how do i mak eit where i hear myself

[hallow thistle]

On a voice changer that you still not say the name, set "monitor" to your speakers/headphones to hear the program.

[quasi condor]

k the dude is an idiot

[hollow gulch]

how can we make Flux-LoRA image models?

[tacit moth]

Can someone tell me what the default block size,extra time and crossfade duration should be for vonovox

[nova willow]

need help with getting the vac (or virtual audio cable) working for my microphone so I can use it for fl studio

[knotty marten]

i got question

[oak falcon]

I want to do collaboration in ml project so if anyone interested so please dm me

[crystal barn]

How to make an AI cover of a song in a certain style? Like a country song as a k-pop song.

I saw one but can't link probably due to rules, but I really wanna mimic however they accomplished it with a different style, idk the software they used

[humble barn]

im looking for that patapon voice model

[wind cosmos]

i fixed my issues awhile ago, but for future reference if you get asked the same question ask them what directory they have their files in, they should have it in their main drive (usually C) and not in downloads

[hallow thistle]

[hallow thistle]

You might wanna look for one in #1175430844685484042. If the voice model doesn't exist, make a request in #1159289738314919936.

[hallow thistle]

Since Vonovox is made to give better audio quality in mind, even the default settings (extra time, block size) would still work fine. For crossfade duration, try set it to 0.15.

[hallow thistle]

A Stable Diffusion Flux.1 LoRA model can be trained just like any other Stable Diffusion image model version. While training 1.5/XL models might be simpler, Flux models would likely involve even more complex dataset settings.

[elder coral]

how many epochs would i train

[hallow thistle]

Around 300 - 400 for a typical RVC voice model. The more epoch number doesn't always mean better model quality, often times have to happen with dataset audio, "pretrained voice model", batch size and settings.

[elder coral]

i'm training an ali g model for about 330

[quasi condor]

or do 10000 epochs and if it overtrains stop training

[hallow thistle]

[quasi condor]

that's what i did

that's why my models is sometimes a thousand epochs

[winter radish]

yo i cant connect it to discord

[wind cosmos]

use vbcable

[winter radish]

i dont see the option

[thick copper]

Is there no voicemod like software which doesn't require installing VBA's

[hallow thistle]

https://cdn.discordapp.com/attachments/1159290139609137264/1460567668582125761/image.png?ex=699b7664&is=699a24e4&hm=318b502ff68c2bf949175429f8fb0b2e9b37c1eeae7da4d1b409b605e89ab3a5&

This might give you an idea.

[thick copper]

anyone

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[thick copper]

Are these illegal

I'm not sure if voicemodulating realtime is illegal

probably not

[hallow thistle]

Not always illegal if you use either voice changer for fun. But for voice models that are trained from copyrighted materials, it's still legally questionable and hard to conclude.

[winter radish]

where do you see this

[hallow thistle]

In my screenshot, it's "Tg Develop's W-Okada voice changer fork". Basically another better working voice changer. https://cdn.discordapp.com/attachments/1159290139609137264/1467527432457424969/image.png?ex=699bbbad&is=699a6a2d&hm=05ffe25c9b09f967aefacf95baa9ee74c9681336b8f674a8066933bf80586264&

[winter radish]

i connected my voice changer but its just mute

[hallow thistle]

Make sure to read help guidelines from here before start asking anything.

[quasi condor]

poor laptop or pc running okada

locally

[winter radish]

yo im having an issue i downloaded the virtual audio cable and im not getting any output from it when i use the output on discrod

[hallow thistle]

How am I supposed to diagnose?

What is your PC GPU? And did you follow any tutorial or guide before?

[winter radish]

yea but i got lost and im on laptop its a hp victus and my gpu is a geforce rtx 4070

[oak rapids]

hey guys i need help with something

[hallow thistle]

[loud rivet]

hello can someone make an rvc of Japanese toji zenin and Japanese chose kamo??

[elder coral]

my model took about 2 hours to train and i clicked on overtraining_detector

[hallow thistle]

There's #1159289738314919936 for requesting RVC voice models.

[viral mason]

No need for that toggle, when training just test each epoch until it sounds good

[tacit moth]

on vonovox how can i make it so i can hear myself?

[hallow thistle]

https://cdn.discordapp.com/attachments/1159290139609137264/1467528034805485742/image.png?ex=699bbc3c&is=699a6abc&hm=55131b72d1dbce534d76faf043b3b72810968a862b9d5e21bf33cec25beb2ab9&

[tacit moth]

so no way rn how i can have the input going to the game and being able to hear it at the same time?

[hallow thistle]

W-Okada voice changer lets you to output to Line 1 and your speakers/headphones (if set to monitor) at the time, but Vonovox doesn't seem to be it.

[tacit moth]

do u think its worth having them both installed only to hear my voice?

[elder coral]

go to my forum

[tacit moth]

where do i get the real time voice changer client at?

[hallow thistle]

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

[waxen epoch]

can anyone make a rvc model for me.. i got a clip thats 40 minutes long.. but theres keyboard noice and mouse clicking noise..

[hallow thistle]

There's #1159289738314919936 to make a voice model request.

[hollow gulch]

where can we use the image models now after Weights gg shuts down?

[unborn tiger]

hello
anyone knows prolific ai?

[winter radish]

i set my audio to the virtual cable and its not picking up anything

[hardy yew]

can you share a screenshot of your configuration?

[winter radish]

my audio configs?

[hardy yew]

from the voicechanger first

[winter radish]

alr

[winter radish]

do u want me to send it in dms i cant send pictures here

[hardy yew]

oh, i thought pictures can be sent here. Perhaps not.
You can send a DM if you want

[winter radish]

alr

[quaint forge]

what can you do

[old crater]

how to increase the performance?
cuz sometiems it stutters

[white edge]

-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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[gentle epoch]

wheres the download link for Tg Develops on amd radeon gpu

[viral mason]

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

Here ya go

[glossy zinc]

the voice changer is gilhch

@proven oasis

[viral mason]

<@&1159293204038955078> is who u should at for help

lol

[viral mason]

this is some random dude

[glossy zinc]

ohh '

mb

[viral mason]

ur good

[torpid jackal]

can someone help me with my voice changer? its refusing to take any files

[rare siren]

Still no answer man?

[tame oracle]

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[tame oracle]

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[quasi condor]

we gave up cuz i don't think u understand anything

like why do u want the old one so badly

[viral mason]

the old one is yucky

tg and vonovox are peak

[quasi condor]

we suggested you to use the new ones but you refused

[rare siren]

I dont want old one anymore

Jst

How do i get cloud one man

[gleaming flame]

what is vonovox?

is w-okada good?

[viral mason]

it's the number one free realtime voice changer

only works for nvidia tho

[viral mason]

yep!

[gleaming flame]

got it

i have it now

tested it in discord

[viral mason]

Promoting isn't allowed:0

[fresh cairn]

yea i am using Invoke right now on linux, and the voice changer been broken for the AMD linux version. (something during the install but haven't seen updateds since last time)

[floral quiver]

i need help

i cant hear my ai thing

@tame oracle can you help me/

[tame oracle]

i can't right now cuz its 1:43 AM and im about to go to sleep

but uh

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[floral quiver]

ohh its finee dw about it

[tame oracle]

I appreciate it

also

dont ping psecific ppl please

[floral quiver]

Yeha ofc

[tame oracle]

just ping the helper role once

:)

[floral quiver]

OHH DIDNT KNOW SORRY!

[tame oracle]

It's okay

^<^

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[floral quiver]

can you by any chance help me

i find it hard to ask the thing

[low shard]

If you elaborate and I'm able to help with your issue then sure, but I don't know your issue, which is why I sent that message that helps users how to elaborate their problems

[floral quiver]

my rvc is not working since i cant hear it on my browser i have it on the browser

i think

i destroyed or deleted somethign essential

[floral quiver]

oh wait i wasnt allowed to tag mb

[rare siren]

soo can u help me get the cloud one?

[hard gale]

You can try huggingface or replicate for most image models. Some smaller projects are popping up too, but those two are the safest bets for now.

[old crater]

gpu : rtx 3050 i5 10400f

how to reduce stutters on voice changer
because i tried multiple ones and there are still some stutters

[gaunt falcon]

dose anyone knows how to fix Something went wrong
Connection errored out. ?

[simple ore]

that happens when you kill the sever process

or you need to provide more details

[gaunt falcon]

i can't place picture here

I opened go-web.bat and selected the sound and model as usual.

after i press Convert
<All keys matched successfully >
loading rmvpe model

\RVC\Mangio-RVC-v23.7.0_INFER\Magio-RVC-V23.7.0>pause
Press any key to continue . . .

[viral mason]

holy old

you might be using something realllly ancient

is this a voice changer or is this applio?

[gaunt falcon]

Do I need to get an update?

ok my holy 30mbps slow ass internet gonna catch up soon

[viral mason]

I need to know if it's one of these two things before I can say anything

[old crater]

how do i fix this?

[viral mason]

what voice changer do u have, did u use a youtube tutorial to download it

[gaunt falcon]

Where can I find the new one?

[old crater]

no
i use w-okada tg fork

[viral mason]

I have my settings like this but I have a 5070ti so u could try a similar chunk size

[low shard]

RVC means Retrieval-based-Voice-Conversion, not Realtime Voice Changer

please elaborate your PC OS, GPU and what tutorial you're using

I found your message: #✦│chat message

Ilaria RVC hasn't been working for a while, if you elaborate your PC GPU and OS I could help you find alternatives

[proud belfry]

My i3 notebook, which can't even run its own operating system, wouldn't be able to handle it, but thank you for your help, please come back Ilaria 😭😭😭

[low shard]

If your notebook runs a browser like Google Chrome it can use Cloud, just like Ilaria RVC did, using remote connected servers with free limits

tbh I don't think Ilaria RVC is coming back (I haven't personally heard any news about it since months)

you wanna try cloud alternatives or ?

[proud belfry]

I tried it once, a long time ago, but something kept breaking.

[junior oyster]

my vc (voice changer) is really laggy/delayed and it is detecting a lot of background noise

[viral mason]

did you get it off of a yt tutorial?

[junior oyster]

yes

[viral mason]

it's old and outdated then

what graphics card do you have? (Nvidia or AMD)

[junior oyster]

amd

[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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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]

this is a virtual cable to use the voice changer in games or discord (Like vb cable) https://software.muzychenko.net/freeware/vac470lite.zip

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

[junior oyster]

wait i had an old vers

[viral mason]

wdym?

[junior oyster]

of the vc

[viral mason]

yea the one u have from the yt video is outdated

[misty cradle]

hello?

guys i have intel hd 520 i7660u laptop can i use the app?

[craggy bough]

no

[fierce falcon]

has anyone used the deiteris wokoda fork with remote play to vc on ps5? i want to use kaggle too because my pc is meek. but im not even sure if this setup can work out or cause too much latency

[whole yoke]

yo i have a RTX 5050 16gb , how do i get/use the app? and which should i use.

[tame oracle]

the RTX 5050 comes in a 16gb variation?

[signal vapor]

ye

[tame oracle]

never heard of a 5050 with 16gb of VRAM

pretty sure it only has 8gb

[hallow thistle]

How do you know if that one "NVIDIA GeForce RTX 5050" comes with a 16GB variant? The only known RTX 50 series GPU that feature 16GB VRAM are RTX 5060 Ti (16GB), RTX 5070 Ti and RTX 5080.

[tame oracle]

[whole yoke]

wym

am i slow?

its 2 8gbs tho

[tame oracle]

[whole yoke]

ive got a computer

[tame oracle]

Mkay

but like

[hallow thistle]

Neither single GeForce RTX xx50 GPU in any RTX or GTX series features VRAM over 8GB. If you mean something else, just explain.

[tame oracle]

a 5050 doesn't ave 16gb of VRAM

[whole yoke]

brb

gpu

or imean

ddr4

ohhh

i have

a

5500

not 5050

[hallow thistle]

To check your PC specs, open Task Manager.

[whole yoke]

im uasihng the vononox

or whatveer that ias

it seems to be working good

but yer ima check

how i check my specs

mb im a pc noob

been mac all my life

[hallow thistle]

On Windows 11, open Task Manager, go to Performance tab.

[hallow thistle]

W-Okada voice changer would only use CPU to process instead, so less likely.

[edgy wasp]

Whats the best settings? I have an RTX 5060

[low shard]

This is a general ai server, please elaborate

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[hallow thistle]

What settings? Is it about games settings or a voice changer settings?

[reef isle]

can u pls help i got a good gpu and cpu even the model and the quailty is so bad idkk y
and getting cuted

[hallow thistle]

Make sure to explain about which program you're looking for, and read help guidelines before start asking anything back.

[reef isle]

its the voice changer

[hallow thistle]

W-Okada voice changer or Vonovox? What is your PC GPU? And were you following any tutorial or guide?

[reef isle]

its RTX3060 12gb

i did follow

but the quailty isnt the same as the video

[hallow thistle]

[reef isle]

and also i used the same model

W-Okada voice changer

[hallow thistle]

You better try Vonovox (the better working alternative voice changer) or Tg Develop's W-Okada fork, instead of the one you got from tutorial on YouTube.

-realtime

[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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[reef isle]

is Vonovox better?

i need a better quality

[hallow thistle]

Vonovox is made to give better audio quality in mind, whereas W-Okada fork needs you to set "Extra" value to 2.7 s. For "too deep" comparison between these voice changers otherwise, I'm not sure how do that.

[obtuse summit]

Can anyone help me in creating a rvc model? i have the last version of applio
specs: nvidia rtx5060 with a ryzen5 5600x and 32gb ddr4 ram

[reef isle]

i tried W-Okada the sound was so bad

[hallow thistle]

The "latest" Applio RVC in question is 3.6.2.

[obtuse summit]

oh i have 3.6.1 do i need to download it?

[hallow thistle]

Tg Develop/Deiteris W-Okada forks and "original" W-Okada are two different versions made by different authors.

This is Tg Develop's W-Okada realtime voice changer fork. https://cdn.discordapp.com/attachments/1159290139609137264/1467527432457424969/image.png?ex=699e5ead&is=699d0d2d&hm=8cd513294f6eee1542b09e7ffbc5e7e21d3694b5ae0fadec57b6bc25b6a20b02&

This is Deiteris' W-Okada fork. While this one has similar UI to the original W-Okada, it's a different implementation. https://cdn.discordapp.com/attachments/1159290139609137264/1467535508019347671/image.png?ex=699e6632&is=699d14b2&hm=eefccdd19fa5eabd67851effecca2698a97a7adc72ccc04646a24d6276a78002&

[hallow thistle]

Not always required or needed, but recommended.

[reef isle]

im using the voice changer demo of this one

[obtuse summit]

ok let me download it, one sec

[hallow thistle]

"This" is the actual original W-Okada version. I'm not sure what to explain about this. https://i.ytimg.com/vi/xwdbwbtO-FQ/maxresdefault.jpg

[obtuse summit]

my dataset is 3:02 min long

[reef isle]

yes the sound here was too bad honstely i treid a good model and a good settings and it still the same

[obtuse summit]

is it good?

[hallow thistle]

Well, I'm not sure what to dispute about W-Okada then. Just try Vonovox at this point.

[reef isle]

will Vonovox work good ?

[hallow thistle]

Better not ask too many questions. I'm more of telling good-working settings.

[low shard]

https://tenor.com/view/n-approved-murderdrones-n-suspicious-gif-7783997419247641730

[hallow thistle]

This is what Vonovox looks like. https://cdn.discordapp.com/attachments/1159290139609137264/1446358776587489351/image.png?ex=699e7b94&is=699d2a14&hm=e3afdb4e05fb6baf89bf79a5551bce5127c3698f55d102d05e10712c5a4ce90b&

[low shard]

would be best to get as clear and as much time as you can

[obtuse summit]

i only recorded all voicelines from N in the episode 1 and made it as clear as possible, and converted it to a wav with audacity to boost quality too

[reef isle]

how can i get it

[obtuse summit]

i finished downloading, now imma exctract

[low shard]

What’s your PC OS and GPU? I can suggest you the right realtime voice changers, the quality is based on different factors but ofc don’t expect to scream with the models

[obtuse summit]

i have w11 home btw, i forgot to say that

[hallow thistle]

I just.

Here #✨│ai-help message.

[reef isle]

RTX 3060 i5 gen 12

[hallow thistle]

Or the more direct way is download Vonovox from this link. https://github.com/dr87/Vonovox/archive/refs/tags/v1.6.9.zip

[low shard]

You might get better results with more voicelines from other eps

Btw converting an mp3 to wav doesn’t boost the audio quality, since it already lost its quality
Be sure to get the highest audio quality you can, using the Amazon Prime serie version could have higher quality compared to the YouTube one since YouTube compresses videos

[obtuse summit]

i did use amazon cuz i need it in italian

[low shard]

I’m guessing you’re on Windows 11 right?

[reef isle]

ye

[low shard]

Lmao didn’t see another italian here since a while, buongiorno

[hallow thistle]

Is that one audio has 32000, 44100 or 48000 Hz? Make sure to check the audio content frequency using program "Spek".

[obtuse summit]

hol up lemme check

[reef isle]

so what should i use?

[low shard]

-realtime

[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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[reef isle]

cuz i used it with a good model and good settings i got a bad result

[low shard]

You can use either Vonovox or Wokada Tg-develop. Vonovox might be better for your setup so i’d suggest to try that first

[low shard]

did you already try vonovox?

[reef isle]

no

[low shard]

try it, also remember to play with the settings and try different models, remember that RVC v2 models are limited and you can’t expect things like a perfect laugh

[hallow thistle]

Let's get it started.

[reef isle]

i used W-Okada

the white one

[obtuse summit]

im using applio to check it and it said this
Sample Rate: 22050
Duration: 3.04 minutes
Number of Samples: 4015872
Bits per Sample: 48000
Channels: Mono (1)

[hallow thistle]

Files of Deiteris' W-Okada look like this.

[low shard]

i’m guessing original wokada from a youtube tutorial? Youtube video tutorials are pretty outdated, don’t follow them

[hallow thistle]

That just shows basic information from your audio file.

[reef isle]

yeee

i will try vonovox

[obtuse summit]

i think its 4800 but im not sure, whats the best quality so i can remake it at the best one?

[hallow thistle]

This is how you check your actual audio content frequency, regardless of which sample rate your audio file stated.

[obtuse summit]

oh its 4800 mine so is it good?

[hallow thistle]

If the actual audio frequency peakes at 16 kHz, it's 16 kHz (doubled as 32 kHz).

[obtuse summit]

ok let me check that

how do i send screenshots?

https://imgur.com/a/dBzxfrB

ill use this

[obtuse summit]

so is it good or not?

[obtuse summit]

this is the screenshot

[hallow thistle]

Yes, the actual audio content from your file actually peaks at 16000 Hz. But beyonds 16001 Hz, it's just noisy high freq noises.

[obtuse summit]

so its good?

[hardy yew]

It means you should train with 32kHz setting

[obtuse summit]

ok

[hallow thistle]

It's questionable. You either try export your audio file as 32 kHz instead of 48 kHz to remove unnecessary high freq noises that occur from 16001 Hz onwards.

[obtuse summit]

ok let me convert it to 32000

alr its 32000 now and doesn't go past 32000

https://imgur.com/a/btgl7xN

what do i put now here on applio

[hallow thistle]

According to the Nyquist theorem, exporting your audio file as 32000 Hz, the double of 16000 Hz, is better known to preserve entire 16000 Hz audio content perfectly. The same situation also applies to other sample rates (22050 in 44100, 24000 in 48000 Hz, 48000 in 96000 Hz, so on)

[obtuse summit]

ok so what do i put now to train it
https://imgur.com/a/btgl7xN

[simple ore]

that's not what it says

it is "to sample 16KHz range you need to sample at 32KHz"

[hardy yew]

Yeah, it's the least perfect scenario, an edge case

[obtuse summit]

im not understanding anymore :/

[hallow thistle]

There are many explanations. Better forget what I say, and continue to training part.

[obtuse summit]

https://imgur.com/a/btgl7xN
what do i put here for 3m of dataset

[hardy yew]

3min is very little, wouldn't expect it to turn out great

[simple ore]

regarding to RVC, if you have audio only going to 15-16k in Spek, you train at 32k

[hardy yew]

Anyway, have you tried reading the training basics in AI hub docs? They're not perfect, but seems to me better than asking about every single step

[obtuse summit]

its at 15k so its good

i did not

[hardy yew]

There's some general info there which might be useful for a starter

[hallow thistle]

Set "batch size" at 4. If your system has plenty of hard drive space. leave "save every epoch" at 10 by default, because this one will make Applio RVC to save models every "10 epochs". Aside from "save every epoch", try set "total epoch" at 300 or 400; this will result in 30 or 40 model files being saved. The higher epoch number, the better the model learned from dataset at each epoch.

I know I'm terrible at explaining, someone would correct me, but anyway.

[obtuse summit]

should i do 400 and save every 20? or 10 is better

[simple ore]

200 and every 10

test the saved models after 50

[hardy yew]

BTW might want to check "save only last checkpoint" or whatever it was called, no need to save those every time and they are quite heavy

[obtuse summit]

i have 1tb so its not a problem, ill do 200 every 10 then

[hardy yew]

(it's in advanced)

[simple ore]

that's for D/G training weights weights

[hardy yew]

Yes, that's why I think it's redundant to keep those

Other than the latest ones

But with 1TB of free space it's a non-issue anyway

[obtuse summit]

actually its 2tb cuz i have another drive

should i put overtraining detection?

[hardy yew]

Nope

[obtuse summit]

k

[hardy yew]

No need to touch any other setting AFAIR

[obtuse summit]

ok i removed the pretrained check cuz its a new model and i did not use a pretrained

[hardy yew]

Check it, you want to use a pretrain

[obtuse summit]

oh

uhh how do i stop it

[hallow thistle]

I'm more of into basic settings, but I feel like no one agrees with me, so while certain settings do work at my own, it might not always work for others.

[hardy yew]

No one really disagreed with your guidelines above, Noobies only said to use lower epoch value at start, but the actual training settings are the same

[obtuse summit]

im new to this so im sorry if i ask for everything

[hardy yew]

Yeah, no worries, the help channel is for this very purpose. Although, as mentioned before, there's usually some sources that can be checked in the first places for the basics

https://docs.aihub.gg/rvc/resources/training/
This is what I was referring to BTW, it's far from perfect so don't consider it 100% truth but it contains some nice general information

[obtuse summit]

• Running on local URL: http://127.0.0.1:6969
• To create a public link, set share=True in launch().
  Starting preprocess with 12 processes...
  100%|████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:05<00:00, 5.08s/it]
  Preprocess completed in 5.09 seconds on 00:03:02 seconds of audio.
  Starting preprocess with 12 processes...
  100%|████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:05<00:00, 5.07s/it]
  Preprocess completed in 5.08 seconds on 00:03:02 seconds of audio.
  Starting pitch extraction on cuda:0 using rmvpe...
  100%|██████████████████████████████████████████████████████████████████████████████████| 71/71 [00:03<00:00, 20.09it/s]
  Pitch extraction completed in 9.54 seconds.
  Starting embedding extraction with 12 cores on cuda:0...
  100%|██████████████████████████████████████████████████████████████████████████████████| 71/71 [00:01<00:00, 37.16it/s]
  Embedding extraction completed in 7.75 seconds.
  Saved index file 'C:\Users\liamv\Downloads\ApplioV3.6.2\logs\N\N.index'
  Initializing the generator with 109 speakers.
  Using HiFi-GAN vocoder
  Using AdamW optimizer
  Using Single-Scale Mel loss function
  Using Float16 for training.
  Starting training...
  Loaded pretrained (G) 'rvc\models\pretraineds\hifi-gan\f0G32k.pth'
  Loaded pretrained (D) 'rvc\models\pretraineds\hifi-gan\f0D32k.pth'
  Using contentvec reference set for validation
  is it going?

[hardy yew]

(and tensorboard is not very useful)

Looks like it's warming up, you'll see a progress bar for each epoch once it's actually started training

[hallow thistle]

Because certain models I trained usually work best at around 300 - 400 epochs, it's why I said like that. You can still follow the guide as your basic reference, by the way, unless you want more perfect audio quality as possible.

[obtuse summit]

N | epoch=1 | step=20 | time=14:46:54 | training_speed=0:00:10
N | epoch=2 | step=40 | time=14:47:00 | training_speed=0:00:05 | lowest_value=28.713 (epoch 2 and step 28)
N | epoch=3 | step=60 | time=14:47:06 | training_speed=0:00:06 | lowest_value=28.713 (epoch 2 and step 28)
N | epoch=4 | step=80 | time=14:47:12 | training_speed=0:00:05 | lowest_value=25.369 (epoch 4 and step 69)
N | epoch=5 | step=100 | time=14:47:18 | training_speed=0:00:05 | lowest_value=25.369 (epoch 4 and step 69)
N | epoch=6 | step=120 | time=14:47:24 | training_speed=0:00:05 | lowest_value=25.369 (epoch 4 and step 69)
N | epoch=7 | step=140 | time=14:47:29 | training_speed=0:00:05 | lowest_value=24.228 (epoch 7 and step 137)
N | epoch=8 | step=160 | time=14:47:35 | training_speed=0:00:05 | lowest_value=22.724 (epoch 8 and step 153)
80%|█████████████████████████████████████████████████████████████████▌ | 16/20 [00:04<00:01, 3.66it/s] yeah its going, i used gemini for the previous try and he said that the lowest value makes it better when it descends

[hardy yew]

This heavily depends on the data, it's quality, it's length and a bunch of other things.
I don't really know what'd be recommended for N's case of a tiny dataset. But it trains absurdly fast because of its size, so training for 400 or even more epochs and then comparing against earlier checkpoints sounds like a nice start

[obtuse summit]

i have a lot of time so no problem if i need to train again

[hardy yew]

I didn't comment on the actual epoch count before as I don't have an opinion on this myself in this case

[hardy yew]

It's 6s per epoch, so time's not much of an issue anyway, you can play around with it easily

[obtuse summit]

k

what are G and D

[hardy yew]

Generator and discriminator, it's from GAN terminology

[obtuse summit]

k

im at 90/200 now

[hallow thistle]

Training a voice model from a 16-second dataset audio would sure be fast to train, which has occured to my "Killer Fish" model, but then anything that's minute long would make RVC to train slightly slower for every epoch, despite the same training settings.

[obtuse summit]

after it finishes do i use applio, w-okada or vonovox

[hardy yew]

You can test it on audio files in applio. Will want to do it on various checkpoints

[obtuse summit]

ok

[hardy yew]

In the inference tab

[hallow thistle]

Use Applio RVC to test your voice models first before going to W-Okada or Vonovox.

[obtuse summit]

imma leave it train and imma afk for a sec

back

N | epoch=126 | step=2520 | time=14:59:06 | training_speed=0:00:05 | lowest_value=13.651 (epoch 58 and step 1159)

[hardy yew]

Each epoch trains over the entire dataset in batches, so the training time will grow more or less linearly with it, not including data transfer time, CUDA initialization etc.

You trained a model on 16s data? Did it actually turn out good?

[obtuse summit]

this is my lowest lowest value, last time i was at 35 now 13

[hardy yew]

Wouldn't pay too much attention to it

[hallow thistle]

This is what I usually do each time I train a voice model in Applio RVC within Kaggle. Because Kaggle provides a limited disk space for each session, it's why I set "Save Every Epoch" to 50 to save up disk space, as well as storing some other model projects there. For training locally otherwise (at least 1TB storage), I'd set "Save Every Epoch" to 10, sometimes with same "Total Epoch" value.

[hardy yew]

Listening to the model gives much more reliable feedback than the losses

[obtuse summit]

i disabled the save only latest so i have checkpoints for g and d too

[hallow thistle]

Turned out good at 400 epochs. The same model at 200 epochs felt too undertrained.

[hardy yew]

I doubt you'd need those, IMO it's a waste of space, but as mentioned previously, it's definitely not an issue for you

Usually if you want to continue training, you want to continue from the last epoch anyway

[obtuse summit]

k

maybe the last doesn't sound good so i did that for convenience

[hallow thistle]

This will make Applio RVC to save all too-large G and D files at every 20 epoch, resulting in significant disk storage usage.

[obtuse summit]

last 5 epoches

alr it finished

Lowest generator loss: 13.651 at epoch 58, step 1159

[hardy yew]

Curious, I need to check what that killer fish sounds like. Is it some more-or-less clear, genuine voice or something heavily processed/artificial? I've noticed that the latter favors longer training and wonder if that's the case

[hardy yew]

Test how it sounds at 50, 100, 200, 400, then can narrow it down if some particular spot seems to sound much better

[obtuse summit]

alr so 200 (the last) is good but my microphone was going bad so idk if its my mic or the model

can i do TTS ?

[hardy yew]

You can, but IMO the best way is to run it on audio files

You can make a sample and infer the models on it and compare the results

[hallow thistle]

"Killer Fish from San Diego" is a meme voice model trained from a meme audio. It's not always high-quality like some other models because the original material was like recorded from a 2000-era phone camera. I didn't have to artificially resample or upscale it, the results model audio (and accent) are almost or more accurate to the original audio anyway.

[hardy yew]

If the last one is clearly the best, it's a hint that perhaps you should train for longer

[rotund zephyr]

yo what does it mean when a model is RMVPE does that mean i cant use it in RVC?

[hardy yew]

RMVPE is a pitch extraction method

It's part of the RVC pipeline

And also the default one usually

[obtuse summit]

200 sounds good but with a wavy voice

[hallow thistle]

Rather than using your microphone, the better bet is to use "a vocal audio file" to do "inference" or test your models on Applio RVC.

[rotund zephyr]

ok just wanted to make sure its not the model cuz i cant hear myself in kaggle

even tho i set my monitor to my speaker

[obtuse summit]

alr so 200 is the best i guess but just sounds wavy

[rotund zephyr]

this is my settings btw if u guys ask for it

[obtuse summit]

oh it was my mic again :/

[rotund zephyr]

cuz idk why i cant hear myself

my mic works fine

[obtuse summit]

do u have fxsound open?

[hardy yew]

Ooh yeah this makes sense. Old recording, presumably low quality and distortions, I feel like those are actually easy for RVC to replicate in some scenarios. And also why longer training is favorable. Thanks for sharing

[rotund zephyr]

yeah

[obtuse summit]

then idk

[rotund zephyr]

can anyone help with this?

[obtuse summit]

try putting it on your headphones instead of fxsound

[rotund zephyr]

doesnt work either

imma try closing fxsound itself

[hallow thistle]

"RMVPE" is an F0 (pitch extraction) model. It's commonly bundled in nearly every RVC (retrieval-based voice conversion) program, as well as realtime voice changer that uses RVC voice model.

[hallow thistle]

That's just audio devices.

[rotund zephyr]

i still cant hear myself, damn

[hallow thistle]

Huh, you still using Opera GX or something?

[rotund zephyr]

chrome this time

ditched the gx for this one

[hallow thistle]

If you use "ngrok" tunnel on this one, make sure to check your microphone permission of your current ngrok website tab.

[rotund zephyr]

ohhh damn maybe thats it

lemme try that then

[hallow thistle]

Also try check your volume settings.

[rotund zephyr]

this should do it rite

nope still cant hear myself gaaaaaaaaahhh\

[hallow thistle]

Yes. Then you might check this one as well.

[timid mirage]

chat, was suggested to change epoch to 190 on a model. how tf do i do that on okada and vonovox

And also, models I’m using are sounding robotic

[rotund zephyr]

yeha man i still cant hear myself lol

[hallow thistle]

Can you send your screenshot of your full W-Okada interface?

[rotund zephyr]

this one u mean?

[hallow thistle]

Click stop button and try set these settings.

How many speakers device plugged to your PC? If that could be the case.

[timid mirage]

can you send me the updated one

or link me it

[rotund zephyr]

just 1 headset

[rotund zephyr]

of my interface?

[timid mirage]

yeah, mine doesn’t even include some of the stuff on there

[rotund zephyr]

how do u change pitch extraction

nvm got it

i did everything, still not workin

[rotund zephyr]

ok finally worked idk why setting line 1 virtual audio as my mic worked but hey its working lol

[viral mason]

Is this wokada deiteris?

If not it's old and outdated

[viral mason]

What gpu do u have?

[timid mirage]

4090 ti

[viral mason]

You should get wokada tg fork, it's better than whatever Eric has

I'll get the links

-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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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]

virtual cable (for connecting the voice changer with discord or games) https://software.muzychenko.net/freeware/vac470lite.zip

Main voice changer file https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-cuda.zip.001

secondary file for vc https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-cuda.zip.002

you'll need these three things, for vac47 after extraction run setup64

[wraith shale]

guys

when i talk

theres like echo

they can hear the voice changer

[viral mason]

You need to do <@&1159293204038955078> when asking for help btw, more likely you'll get help

[tame oracle]

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[slate fractal]

где русские женские голоса

[sullen raven]

Hey I am looking for a way to use AI to convert/translate a SUB anime episode into a DUB anime episode on my computer locally. What is the best way to do that?
OS: Aurora Linux
GPU: Intel Arc A770 16GB
CPU: Intel Core i5-12400F
Memory: 32GB DDR4
Storage: 1TB

[sweet dock]

Does anyone know why whenever I try to add a background track it doesn't work?

[fierce falcon]

hey im on kaggle notebook editor, how exactly do i run "cells"? do i need to copy and paste the python somewhere

using deiteris w okada btw

[sweet dock]

Forget what I said, i was just being to hasty lmao.

[fierce falcon]

nvm found it out!

[olive lintel]

hey, i'm kinda new to these stuff. how do i install a realtime voice changer? and are those things safe?

[hardy yew]

-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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[hardy yew]

You have guides here

Assuming you have an Nvidia GPU, use tg develop or vonovox

[tame oracle]

A NVIDIA GPU is only needed for Vonovox

[hardy yew]

It's a preferred hardware no matter the app, really

But tg develop has an AMD variant

[tame oracle]

yes

[tame oracle]

Vonovox doesn't have DML version meaning you can only use NVIDIA GPU's

[hardy yew]

I'm very well aware

[tame oracle]

Mkay

if you say so

[viral mason]

U can use either Vonovox or Wokada tg fork, both are great choices although Vonovox like they both said is Nvidia exclusive

Anyways I'm gonna go back to crying in my bed

[cunning mist]

which one does best on like a lower end gpu

[tame oracle]

Real

[viral mason]

How low are we talking? If it's a smaller Nvidia card I've heard Vonovox still can run on them, but tg fork unsure even tho it should technically be less intensive on either AMD or Nvidia card

[cunning mist]

its a gtx 1660 ti mobile version

[viral mason]

I've seen my friend run tg fork on a laptop amd GPU somewhat alright

[viral mason]

I.. dunno what that means

[cunning mist]

like the lptop one

[viral mason]

Oh, I saw mobile and thought u were on a phone or smth

[hardy yew]

The GPU support is architecture/torch version dependent in general, so both apps should behave similarly in that context

[cunning mist]

oh lol

[hardy yew]

(and also I've run my 1070 in both apps, I think 1660ti is fine, even the mobile one, as AFAIR it's a newer model anyway?)

[viral mason]

But I've heard Vonovox can possibly run on a regular 1660ti I think

Capy knows more

Smart creature

[cunning mist]

i think my version runs like around the same as a 3050 desktop

[viral mason]

Oo

If it's like a 3050 it should work just fine with no issues

[cunning mist]

3050 performs 16 percent better

kk

[viral mason]

I am sleep deprived btw so don't take everything I say as fact, I don't even trust myself rn

[cunning mist]

i probably will upgrade to a desk top pc soon

[viral mason]

Trust me a 5070ti or higher will change your life forever

[cunning mist]

i was looking at the 5070ti

[hardy yew]

I've heard of weird performance shenanigans in terms of various GPUs, someone had a better performance on 2070 than 4070 or something like that. Anyway, as long as the GPU is supported and there's enough memory, the performance is not going to be crazy different between various models. Most delay comes from the audio pipeline anyway and GPU has no impact on that

[cunning mist]

or the rx 9070

[hardy yew]

(also locking GPU clocks changes the behavior massively sometimes)

[cunning mist]

there lowkey a good deal

[hardy yew]

(still, hardly ever needed)

[viral mason]

AMD is cool but you'll lack the ability to run Vonovox

[cunning mist]

but im prob gonna wait for the ram storage

[cunning mist]

yeah

i think amd is more price ot performane

mance

[viral mason]

Perfomane :3

[cunning mist]

🙁

[viral mason]

Sorry I thought it was funny

[cunning mist]

lol

[viral mason]

Not making fun of you I just got giggly

[cunning mist]

lol

my friend has like a spare 3060 in his room i might steal it

[viral mason]

But yea definitely wait a bit because prices are wayyyy high

[cunning mist]

fr though

[hardy yew]

I want a friend like that too

[cunning mist]

mmm

[hardy yew]

got more of those laying around? 👀

[cunning mist]

..

[hardy yew]

(GPUs, not friends)

[viral mason]

Gotta wait till the bubble Nvidia has made to pop for prices to drop lower

[viral mason]

I want friends

[cunning mist]

getting a 3060 is still be a massive stepup from a laptop

[viral mason]

For real

[tame oracle]

Laptops with a 5090 in them be like: 💔

[hardy yew]

Might take some time...

[cunning mist]

..

[viral mason]

Open AI more like where did my ram go

[hardy yew]

We're here for you, friend

[viral mason]

Yay

You're my friend now :3

We're having soft tacos later

[tame oracle]

[cunning mist]

get the rtx 6000

[viral mason]

How do they put those big ol cubes in them though..

I have one in my pc and it's biggg

[cunning mist]

1/3 of the size so 1/3 of the performance

[tame oracle]

they just shove it in

[cunning mist]

or 1/2 of performance

[tame oracle]

make the chassis thicc

[viral mason]

Freaky laptop

[tame oracle]

lmao

[cunning mist]

i think its like the size of ddr5 ram

[tame oracle]

[viral mason]

I forgot why I'm here rn

[cunning mist]

real

[viral mason]

Oh yeah

Uh

Good pc

Give them your wallettt

When prices drop

[tame oracle]

Oh good

[cunning mist]

lol

[tame oracle]

i was about to say something before that last part

[viral mason]

Do you have a wallet?

I wanna eat it

[cunning mist]

should i get a pre built or should i make it

[viral mason]

Not the money

[viral mason]

Pre built is easier to get set up

Mine is pre built

It blows and stuff

Glows

Rgb

[cunning mist]

because therees like this store where u pick parts on their website than they make it for u

[hardy yew]

I guess it depends on where you get it from, there's companies that make decent prebuilts and ones that make overpriced garbage

[viral mason]

I got mine from microcenter

Good deal on mine

Only 2000 dollar

[cunning mist]

wait what specs

[hardy yew]

Perhaps the market is also completely different between countries

[tame oracle]

my wallet has spongebob on it 🥀

[tame oracle]

"only"

[viral mason]

Dude.. I need one just like that

[viral mason]

5070ti, and I forgor everything else

[cunning mist]

oh lol

[viral mason]

I'd have to look when I wake up

[hardy yew]

"when I wake up" lol

[tame oracle]

lmao

was 20 dollars

in Puerto Rico

[viral mason]

It's 4 am and I'm not sleeping when I should be

[tame oracle]

its 2 am for me

[viral mason]

°^°

[hardy yew]

Lies, it's 11AM

[tame oracle]

[viral mason]

Everyone's wrong and time is fake

[tame oracle]

okay

[hardy yew]

You just gotta adjust your clocks

[viral mason]

The earth just orbits and stuff and the sun goes away

[tame oracle]

do u guys believe in gravity

[cunning mist]

uhmm

[viral mason]

No Mr nueton made it up

[hardy yew]

It's the sun that orbits earth obviously

[tame oracle]

that's what ive been saying

[tame oracle]

[viral mason]

How else am I stuck to the ground

It's because I want to

Obviously

[hardy yew]

Glue.

[viral mason]

But

If it's glue

Where's the gorilla

[hardy yew]

Winton is in overwat

And in my heart

[viral mason]

Overwatch 2 is now regular overwatch

Yay

They added a kitty cat

[hardy yew]

Overwatch 2 has never been a thing

[viral mason]

Was that overwatch?

Cat

[hardy yew]

Yes

And it's beautiful.

[viral mason]

Meow

[hardy yew]

Is this still AI help?

[viral mason]

There is something severely wrong with me 👅

[viral mason]

We're helping uhm

The cat

The silly looking cat

Eastool

[tame oracle]

[viral mason]

Yeah

I would be such a good helper (not really)

[hardy yew]

https://tenor.com/view/cat-cam-cat-camera-cat-wide-cat-wide-gif-16729570673879006163

[cunning mist]

cat

[mint vault]

Is the voice changer down?

[viral mason]

What

[mint vault]

me and my friend dont have access to open in

[hardy yew]

To open what exactly?

[viral mason]

True what is this voice changer, the only three I know of are Vonovox, wokada tg fork, and wokada deiteris fork

If you guys got it off a yt tutorial it's from ancient rome

And covered in dust

[hollow drift]

Hello guys any idea how to get realtime voice changer to work in cs go fluently ? If anybody got it setup could you guys send it in chat ? The app works perfectly in discord etc but in cs it seems that it’s not picking up the voice/latency seems to matter

[mint vault]

Anything, open the voice changer it just wont load

[hardy yew]

If it's not picking up voice, I'd double check you have the correct audio input in game's settings.
Other than that, do you cap your framerate? If you don't and it squeezes the max from your GPU, there's no spare room for the voicechanger to operate

[hardy yew]

See the message above from Worm, we don't even know what exact app you're trying to use, but chances are it's something very old, in which case you'd want to download tg develop or vonovox anyway

[hollow drift]

I got 4070 ti super and amd ryzen 9 7900x3d so there should be no squeezed from GPU/cs is mainly CPU based game

I have installed virtual audio cable aswell

Is there any specific apps settings I should use ? And I installed real time voice changer/think it’s okada but not sure 100% is it the best one as free version or ?

[hardy yew]

Even if a game is not heavy on the GPU, if the framerate is not capped it will often run at 100%. Anyway, this can be easily verified with task manager or HWINFO64

[hardy yew]

Well, you definitely want the audio cable to be the microphone input in game.
And about the voicechanger app, if you have something old, see tg develop's w-okada or vonovox, both are nice and fresh.

You can see the guides and download links here:

-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 Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options 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

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

⚔️ Wokada Tg-Develop Fork vs Vonovox

For Windows Nvidia, Both Wokada Tg-Develop 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 Tg-Develop 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

[hollow drift]

Thank you goat

[low shard]

elaborate

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[hallow thistle]

If that one "voice changer" is down, it might be the online website version, isn't it? Make sure to read help guidelines before start asking about which voice changer supposed to be.

[mint vault]

any recos for voice changers?

[hallow thistle]

[mint vault]

;-;

[hallow thistle]

What is your PC GPU? And did you follow any tutorial or guide before? These questions are what I really want from you, aside from simple statements.

Check if you have followed any material to install and use the "voice changer" in question. To check your PC GPU (graphics processing unit) name, open Task Manager or press "Ctrl + Shift + Eac", go to Performance tab, spot if there any GPU 0 or GPU 1 there.

[hallow thistle]

W-Okada voice changer is generally a free and open-source program.

[hollow drift]

I am using this one from the description in the video https://youtu.be/SxdnGxicJOg?si=u3bnF9m3e1O6UOp1 / video is called how to easily download and use AI al girl voice changer by novision if you don’t wanna click on the link

I am unsure if it’s okada or not

But is the sound/latency better in w okada ?

[hallow thistle]

There are like three different implementations of "W-Okada voice changer": one is actual original W-Okada (the one you likely got from most tutorial videos on YouTube), one made by Deiteris, and then another made by Tg Develop; the latter two are more recommended over the former one.

There's "Vonovox" the alternative voice changer that made to give better audio quality in mind, doesn't based on "W-Okada" and isn't open-source like others. Vonovox might be another option if you're looking for superior audio quality.

[hollow drift]

I mean I am looking mostly for not such a big delay and so it works rest I can work around 🙂

[hallow thistle]

So, Tg Develop's W-Okada might suit better.

[winged cairn]

Is anyone looking for help with Ai dm me

[hallow thistle]

So what's this channel for then?

[low shard]

please elaborate your PC GPU, OS and what you're using

[hallow thistle]

And also don't always just react one message with a check mark emote to indicate something else. Try say anything here, directly.

I have an explanation about why Spek only shows up audio content to the half sample rate of audio file. I doubt if anyone wants to know about this.

[hollow drift]

Checked on the OKADA 😄 and it’s the one I have installed or atleast from what I can see I am wondering now have anyone tried using it in cs ? As that shit have huge input lag itself @hallow thistle how much should I cap my fps at ? Is it around 20% of the graphic card usage or less

[hallow thistle]

Let's be real. Which "W-Okada"?

"Tg Develop's W-Okada fork" looks like this. https://cdn.discordapp.com/attachments/1159290139609137264/1467527432457424969/image.png?ex=699fb02d&is=699e5ead&hm=b119f194c3c8e1dc39145d6e6935164c16b1a7419b8ab968e24ff26dc2ebd942&

[hardy yew]

Wasn't this already explained? It's fairly simple and, as you mentioned before, is a direct implication of Nyquist sampling theorem

It's not about Spek really

[hallow thistle]

Yes, but what I explained got overshadowed by someone who knows better, so I kind of felt hurt to explain further.

[crimson geyser]

i put my microphone as input and vb cable as output. but i dont hear any sounds i put my microphone as input and vb cable as output. but i dont hear any sounds can some one help?

[hallow thistle]

You know, I didn't mean to provide the same information about voice changer all the time, I'm more of explaining things, even if what I say deemed horrible that anyone could pop up to correct me.

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[crimson geyser]

GPU RX 5700XT
Operating System WINDOW 10
Detailed Description I installed VCC (voice changer) but it doesn't work, i mean i choosed and setted up everything, i put my microphone as input and vb cable as output. but i dont hear any sounds, i choosed vb cable as input in discord and other apps
Tutorial: No
screenshot:

i cant send screenshot

input : My mic
output : Vb cable input

[hallow thistle]

Download and try "Tg Develop's W-Okada voice changer" instead, "https://github.com/tg-develop/voice-changer/releases/download/b2397/voice-changer-windows-amd64-dml.zip" download link. If you ever feel like "VB-Cable" won't work, there's another different program named "Virtual Audio Cable lite".

[crimson geyser]

how can i setting virtual audio cable

i download it

[hallow thistle]

https://docs.aihub.gg/realtime-voice-changer/local/tg-develops-w-okada-fork/#virtual-audio-cable

[crimson geyser]

still not working

idk why

i bought a new microphone, but when I use the voice changer it doesn’t work my old microphone used to work before, but now it doesn’t either.

[hallow thistle]

https://cdn.discordapp.com/attachments/1159290139609137264/1460567668582125761/image.png?ex=69a013a4&is=699ec224&hm=7a8de193253e413619db13503ce97b2c68dc23656540de6899ff28b2d1fd8dae&

On W-Okada, set "monitor" to your speakers or headphones to hear the program; otherwise you won't hear any sound from the program on your speakers.

[twin terrace]

i want arabic girl voice model

[hallow thistle]

Unfortunate thing, it won't be here.

[twin terrace]

where can i find it

[crimson geyser]

still not working

[hallow thistle]

How so? What browser are you using?

[crimson geyser]

chrome

[hallow thistle]

#1175430844685484042

[twin terrace]

i didnt find any arab models

[hallow thistle]

Make sure to allow mic permission to your current voice changer tab.

[hallow thistle]

[crimson geyser]

bro oh my god

thanks you so much

[hallow thistle]

That's one of simple tricks I can think.

[hardy yew]

I understand and am sorry that you felt hurt. I don't think you should, honestly. There's a lot of helpful people around here with different opinions and experience, for me it's more of an opportunity to learn from each other than a reason to feel worse. It sure was nothing personal after all.

I've obviously also been corrected numerous times, but each time it's valuable knowledge which I'm here for actually.

[hollow drift]

@hallow thistle I’ll send screenshot with one I got as I am not home right now

And I can send over cs settings inside of the audio like I stated previously it works everywhere else I read about voice changers in the forum of steam/ it said windows input should be the same with I did + steam settings

But some guides said that every time a game should be opened the input device should be changed back to default one and then changed to VB cable meant OUTPUT not input 😭 saw it now

[hallow thistle]

You still didn't say which "W-Okada" you were refering to.

[subtle cedar]

-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

[subtle cedar]

Since that Weights is gonna discontinue next month so which is Google Colab(s) that for create AI Covers and for the AI Covers one can use music from Google Drive for iPhone or Apple users? (Well I don't even have an iPhone yet but I plan to buy iphone in the future)

[hollow drift]

Ill send picture later when I am home

[subtle cedar]

Okay but I need someone or anyone answer me for this and I'm not even new to ai Covers I just ask which is one of the modern Google Colabs that for AI Covers and because I don't remember if music from Google Drive can use for Google Colabs or Gradios too why I asked about this because I don't use Google Colabs for a long time ✌️

[bold salmon]

hii currently wanted to try RVC from venovox but on precompile section idk which one to download

https://cdn.discordapp.com/attachments/898977516272889856/1476233723941883904/image.png?ex=69a0614a&is=699f0fca&hm=37d54bb2088788ab6bfcbe5821ebef731556ab30159df777137d53528e329bd8&

[hardy yew]

You can just go for 1.6.9, it's the recent release.
There's also beta, but you'd have to download some patches etc, more effort and less stable

Vonovox169.zip

That's the release

[bold salmon]

alright

do i still have to download vonovox from github too or?

[hardy yew]

It should be the same thing. Anyway, you only need one zip, it contains everything

[bold salmon]

as right now im downloading and running setup.bat from github since idk which to zip to downloads before from the huggingface

[bold salmon]

how can i hear my own voice? is it automatically turn on

[hardy yew]

In vonovox there's no built in way to output audio to both virtual audio cable and your speakers.
You can do it in windows audio settings by going to recording tab, going into properties of virtual audio cable and then enabling "listen to this device"

Unless you don't want to use the audio cable, then you can just output to speakers only directly in Vonovox

[viral mason]

Why..

[quasi herald]

Hey, sort of failing at the first hurdle of this which is embarrassing. I have installed the latest version of Applio, and run it via the .bat file, and it seems to be working. I cannot, however, use the run-tensorboard.bat file, it returns the error No module named 'pkg_resources'
I have ran pip install setuptools
which shows that I already had the most up to date version of it, so I am at a loss.

[simple ore]

download the compiled version

you dont need to "install" it

[quasi herald]

I have downloaded the compiled version, install was meant to be more generic in the sense of I have extracted and run it

I have downloaded v3.6.2 from hugging face, which came with the 2 .bat files

using 'run-applio.bat' works perfectly, using 'run-tensorboard.bat' doesn't work as I would hope, returning the error I said above

[humble mulch]

i need some hep for my voice ai agents

[simple ore]

I see, thanks

[simple ore]

unfortunately tensorboard has not been updated to support python 3.12

[quasi herald]

Ah, that'd do it, so tell it to use a different python version. Which is the latest version that tensorboard supports? (And by extension, how do I run tensorboard specifically using a different version of Python?)

[viral mason]

Tensorboard isn't really needed for training tbh, I just train up to around 300-500 epochs and test each one after it hits 100 epochs

Just listening to samples you make is better than any graph

I haven't used tensor in a long time

[simple ore]

you can install a separate python installation of 3.11

then install tensorboard under that

[quasi herald]

Fair, may just freeball it, was just going by the wisdom of the doc, was unsure specifically why it was wanting me to use tensorboard as it would be relatively obvious to tell when it has been overtrained

[viral mason]

If you hear it becoming very robotic or just worse quality that's probably when it's overtrained

[quasi herald]

Thanks :D to both of you*

[simple ore]

tensorboard is useful to see whether there's something weird going on / model exploding / etc

it is not a good guide to tell that the training is done

since the training metrics are only for the model trying to reproduce its own dataset

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[wraith shale]

guys

pls help

when i talk

they can hear its ai'

i mean like

they hear them self

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[hollow drift]

can somebody assist me a little with voice changer settings ?

[hardy yew]

feel free to describe your issue here

[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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[quasi herald]

Is it assumed that I have a single audio file for dataset creation/pre-processing? Or can I upload multiple audio files? I am using Applio if that explicitly matters?

[hardy yew]

the input for your dataset in applio is actually a folder

it can be more than one file

[simple ore]

as long as you dont do something stupid, like uploading random 1-5s chunks and not using preprocess to slice them, it should not matter much

preprocess slices audio into small pieces and they need to be about the same size

so 10x 5minute files is fine, 1x 50minute file is fine, 1000x of 1-10s chunks - not good

[quasi herald]

Gotcha, thank you very much

[quasi herald]

How do you make a .index file for a path file generated via the Voice Blending feature in Applio?

[vague saddle]

yo

wstg

wsg

[quasi herald]

I am trying to use the ElevenLabs TTS Plugin in Applio, but any time I try and use it, I receive the error:
AttributeError: 'ElevenLabs' object has no attribute 'generate'

I have given it an API Key as I thought that may have been the issue, but that hasn't helped at all.

[sullen raven]

Hey I am looking for a way to use AI to convert/translate a SUB anime episode into a DUB anime episode on my computer locally. What is the best way to do that?
OS: Aurora Linux
GPU: Intel Arc A770 16GB
CPU: Intel Core i5-12400F
Memory: 32GB DDR4
Storage: 1TB

[toxic kestrel]

is it possible to get real time voices? if so what specs do i need

[low shard]

please elaborate.

!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 4060 8gb vram desktop)
• 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.

🚫 Prohibited Topics (We Will NOT Help With These)

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

• ANY illegal activities.
• NSFW/Porn.
  Requests for these topics may be ignored, not helped and 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.
• Don't Ask To Ask.

[steel jolt]

point me a dark version of chat cpt

[hardy yew]

ChatCPT isn't real, it can't hurt you

[steel jolt]

really thank you for this but i wasn't asking about the aperiance i just want a free non limited dark gpts like the dark web

[supple cipher]

Dig AI

Can be found on DauntDotLink using TOR browser

Or, jailbreak your own

[fossil crane]

for rvc, what do you think is the best way to use rvc for inferencing alongside a daw?

[feral tundra]

ai

[dense drift]

Its giving me error

When i start processing

[fossil crane]

no, is there an app that can integrate rvc into smth like ableton?

a vst

a max device

so that i don't have to constantly switch

or leave my browser open

[timid arrow]

@fossil cranewhat is rvc

i know how to make vst3s / au

so depending on what it is i might be able to whip it up for you

[viral mason]

VST files have nothing to do with RVC

[timid arrow]

what is rvc

[viral mason]

It's a retrieval-based voice converter software, I would ask a helper tho who might know more as all I know is how to make models and use the real-time voice changer

[viral mason]

There really is no way, I don't see why it would be a bother to switch between either the local version of applio rvc or the browser version

Why do u need it for a daw software anyways?

I use fl studio so I'm curious what you'd be doing

[wind adder]

Cannot register on weights site I have a voice model that I want, what should I do? Models not available on other sites... TT

[hardy yew]

That is the correct question.

For e.g. vocals you'd want to do offline inference on audio files anyway. For realtime... Maybe that would be some use case for a VST, but there's no such thing and I'm pretty sure there won't be. Too much effort for something with very little benefit, something which can also be achieved another way

[viral mason]

What model is it?

I have an account and can download it then give u it I guess

[viral mason]

I do use both fl studio (a daw) with wokada tg in real-time to add extra voice effects to my characters I play to better sound like them while in Vrchat but that is the only use besides adding the effects to those same models just for covers or a song

[wind adder]

I want yushi of NCT WISH TT Can you send it to me? Thank you so much XD

[viral mason]

It would be nice if you could send the link to the model if possible

Makes it easier to find

Btw you can also look here
https://voice-models.com

[hardy yew]

Yeah, that's the potential VST use case I mentioned (realtime). But I think it's better that it stays separate from the DAW anyway, especially that it's prone to various issues

HOWEVER

Whether it could be easily wrapped into a VST... hmm

[wind adder]

https://www.weights.com/models/cm2a1mkw90l7a6lvvtayp4lbb
I've already searched the link you sent me 🥺

[viral mason]

I agree, keeping them separate softwares works much better even if it causes dekay

[viral mason]

Aw alrighty, I'll let you know when I've moved to my pc to help u

[wind adder]

Thank you so much 🥲🫶🏻

[viral mason]

You're welcome ^^

[bold salmon]

when using venovox and joining roblox i notice the effect start to shutter and muffle is this normal or

[viral mason]

Watcha using it on Roblox for..

[fossil crane]

use it to transform my voice into other characters

[viral mason]

That's the same thing I do but I use it in real-time for vrchat :3

[fossil crane]

cool

[bold salmon]

uhmm just little bit of

[viral mason]

https://tenor.com/view/breaking-bad-better-call-saul-mike-ehrmantraut-gif-26551023

of wha

[bold salmon]

of talking

[viral mason]

mmmkay

[bold salmon]

soo you have idea how to fix thed lag?

[viral mason]

what is your pc gpu?

[bold salmon]

gtx 1060 6gb

[viral mason]

that gpu is a little old but should be able to run vonovox fine, there's a new full beta release coming out soon in a couple hours

I'd ask in the official vonovox server for help with that tho

[lapis kiln]

@viral masoni have rtx 4070

What are the best settings I can change or adjust to get the best performance and realistic sound?