laggy and delayed voice any help?

[small sapphire]

my model is laggy and delayed id love some help

[icy orchid]

Before anyone can help you;
can you explain how laggy is laggy. Can you explain how long it takes before you say something and the model reacts ?

For GPU settings, make sure your video card is selected. If it says CPU , (try to) change it to your video card.

Can you tell us what video card is in your computer ?

[rustic bramble]

I don't know if you have other working models or not, but I am going to assume all models are a bit laggy and that from the tag you use W-okada.

To address the issue of the choppy voice and it being delayed let me explain how it works.
In the screenshot you can see my personal settings, now the chunk and Extra both determine the delay, but that will drastically be changed depending on the mode you have picked.

For example let us say we are in the CLIENT setting. This would mean that it takes roughly 3 seconds to speak and and even then it might be choppy since we have barely any chunk despite lots of extra, this you can however avoid by giving it more of both. I usually ran it at 0.34 and 0.35 and it was okay. though you may still need to assign more depending on your computer specs.

IF we however use the SERVER setting, things are a lot different. Chunk determines the delay and extra the load on your computer now, the time it takes with the setting displayed is at most half a second. I found out that switching the two would still result in large delay but this way it worked well. Also depending on what Model you have it may be choppy by nature or even incompatible with the program itself. I ran into that with a model before.

All of that still not helping?

You might run AMD or Mac, the W-Okada if it is the voice changer, works well with NVIDIA since that is what the cuda version works best with, or maybe you have such graphics card like me or better but you never got the cuda version. There is a way to run the program still even if you do not have a good PC, but it takes more work.

If you need a guide I suggest a video, I am not sure if I will be bonked for sharing a link to a youtube guide since I barely post or read things here, but here you go. https://www.youtube.com/watch?v=_JXbvSTGPoo

That video helps with the setup, they also have a guide for weaker PCs and those with no GPU.

[craggy pollen]

that video is extremely outdated, it uses an over year old version of old wokada, don't use yt tuts for voice changers ever

If you use that one, forget about it, it's old, and make your own help post

[craggy pollen]

This is a General AI Server, AI has many fields, so we can't know your issue

Please Elaborate:

• your PC GPU
• your operating system
• what you want to do
• what tutorial link are you using
• a screenshot of the program

[small sapphire]

its AMD

amd radeon

[craggy pollen]

that's integrated graphics, its weak asf

do you have any dedicated gpu?

[craggy pollen]

thats weak asf

[craggy pollen]

Your pc is too Weak for Wokada locally, You got 3 options:

• Buy a better pc
• Run it locally (on ur pc) using the CPU mode of the wokada fork which has better performance https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/ (but this isn't suggested as it could be unstable)
• Use **cloud **(remote good pc):

About Cloud, there are different services:

• Google Colabs (4 hours daily of free T4 gpu, easy to use, require only a google account) :
  • W-Okada's Deiteris' Fork Voice Changer Google Colab (currently works only on google colab PAID tier)
• Kaggles (30 hours weekly of better GPUs, T4x2 & P100, harder to use, requires an account and a phone number):
  • W-Okada's Deiteris' Fork Voice Changer Kaggle (the best and only working one currently for free)

[small sapphire]

ok

[craggy pollen]

let me know

[craggy pollen]

@small sapphire any updates?

[mortal crest]

is there still a cpu mode of wokada?

[echo vessel]

Yes, there is the stripped-down, CPU-only variant of Tg Develop's W-Okada fork. For more help about the voice changer, better ask in #✨│ai-help or create your own thread in #1192011222023950368.