Sound is weird, way too bad

[mortal sluice]

Any voice model keeps sounding bad when i speak + sometimes i speak and it kinda doesnt fully work as in theres no voice, maybe bc it doesnt detect it idk

dk if its bc im using AMD or why

got the AMD Radeon RX 6800

i5

Windows 11

Been 6 hours already here, 3 trying to download w- okada getting multiple help with AI and never fulling installing it now last 3 to install this one just to see that change input or ouput from anything else besides MME freezes the app

[modern cloud]

that's RVC GUI from Mainine RVC, not suggested since like 2 years

dont use video tutorials

delete everything

[modern cloud]

-realtime

[obsidian portal]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

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

• Vonovox

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

⚔️ Wokada Deiteris Fork vs Vonovox

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

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

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

• RVC GUI Mainline Realtime

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

[modern cloud]

get wokada deiteris fork

[mortal sluice]

so its trash

ill try that aight

[mortal sluice]

understanding how to donwload this giving me headaches

[modern cloud]

hello, could u explain me what's wrong in the installation? did u get vac lite?

[mortal sluice]

NVM 😭 i click show that last asset

it was the last

[modern cloud]

why are you downloading from the github?

dont click the first link off the ai hub docs guide in the introduction, it was just for the introduction lol

you should read the guide, and get vac lite first :)

[mortal sluice]

ye reading it rn

[mortal sluice]

kinda confuses me all this stuff at once ngl

what is vac lite

telling me to install it before anything

[modern cloud]

A VAC (Virtual Audio Cable) makes a fake audio device, used to re-route the audio of different programs

In Wokada context, it's used to get the output of wokada as the input in other programs

It's a must

[mortal sluice]

oh, i was using one but if its better

[modern cloud]

if you mean vb audio cable, it's not suggested for windows users, many reported random issues with it, you can uninstall vb audio cable

[mortal sluice]

btw i jumped from here to here bc it says xattr command is wrong idk if this was meant for the users above not amd intel

[mortal sluice]

aight

[modern cloud]

you're following the wrong part, that's for Mac Intel chip users

This is the one for AMD GPU users https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#download-amd-intel-and-cpu-on-windows

[mortal sluice]

no technically i did it fine

deadass im so dumb idk why i showed that on the photo but i did install voice-changer-windows-amd64-dml

extracted and double clicked

though its supposed to open google page and not software?

[modern cloud]

I hope you got it from the link I sent, else if you got the mac intel chip one, your gpu won't be recognized

[modern cloud]

great then :D

[modern cloud]

u extracted it, and not just opened the .exe from the non-extracted folder right?

[modern cloud]

not sure what you mean by google page, but it uses a Web User Interface https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#why-does-it-run-in-a-browser-and-not-its-own-window

[mortal sluice]

yea here but im confused on how his works and how i use my own models

[mortal sluice]

its on google

[modern cloud]

I'm guessing you're saying that it's opening on your default Browser, Chrome, since Google is a Search Engine

so yeah, that's normal

[mortal sluice]

yes

[modern cloud]

extra: 2.7

The difference between Server and Client Audio in Wokada Deiteris Fork:

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

[mortal sluice]

I mean sounds like server is better

Idk

what would you recommend , idm difficulty you adapt to it so just whichever is best ig

[modern cloud]

it really depends on either u want to take more time to read another a bit complex guide for some lower delay or got noise/echo issues and just want the eaisest way

i can't know if you might have echo/noise issues, that's why it's best you decide on your needs/situation

[mortal sluice]

true, havent tested

[modern cloud]

let's use Client for now then
set it to client
input: microphone
output: line 1
monitor: headphones (optionally only if u want to hear urself)

[mortal sluice]

wasapi_

?

wait

nvm

client yes

[modern cloud]

wasapi is only for server

yup

[mortal sluice]

ye mb

[modern cloud]

did u do this?

[mortal sluice]

wait i only got file or system to choose

[modern cloud]

did you give microphone perms in your browser for that page?

[mortal sluice]

oh right

did yea

oh god sounds definitely not like the what ive choosen

already did all, it does sound better as in clear not freezing or broken (well dk)+ but voice is still bad

[mortal sluice]

index: 1 Got: 256 Expected: 768
Please fix either the inputs/outputs or the model.

i get voice errors

or something

all models sound same........

[modern cloud]

try other models, play with the pitch

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

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

[mortal sluice]

I'ma try it tomorrow really , but fp32 barely did much , it's really weird , the voices just sound robotic , so fake
Also the 256 and 768 isn't this for V1 and V2 models

Idk I'm so tired 2 am rn ill try tweaking stuff tmr

Thx

[modern cloud]

lets see tmr

[signal ermine]

Using the original RVC GUI realtime mode in 2025 is a crazy work. There are better alternatives than this one, which one of them being W-Okada.

[mortal sluice]

aight im back, tweaking this stuff but still dont sound much like it

[mortal sluice]

ig, i didnt really know, i had been searching just google videos but seems are all outdated

[modern cloud]

try other models and play with the pitch

not every model is perfect

and RVC can't always do extremely realistic non-speech sounds like screams or laughs

[mortal sluice]

no really i tried alot , good ones too

i feel like i sound nearly the same on all of them

[modern cloud]

can u show a screenshot of ur settings

[mortal sluice]

cant even use v2 of this model

all including men and men to woman sounds idk not so of thatr character

and this 256 or 768 idk wsp but some mdoels i cant use them

or give error

[modern cloud]

RVC v2 models are the architecture they get trained on

it's not a thing you can change nor a setting

most model sin #1175430844685484042 are rvc v2

most, not all, there are some older rvc v1 models

or gpt-sovits models which aren't compatible at all, because they aree TTS not STS

be sure to get RVC v2 models, put extra to 2.7

[mortal sluice]

ye

[mortal sluice]

i think that helped a bit

btw onnx files not compatible?

rn 7.4/16 gb GPU usage

its so surreal this vocies lol

[modern cloud]

you'd need to convert models to onnx yourself via advanced settings,

Convert to ONNX: Reduces delay and slightly reduces gpu usage. Enabling this increases CPU usage by around 5-10%. Reduces the quality of the voice a bit. If you decide to enable this, pair it with rmvpe_onnx for even less delay

[mortal sluice]

OH

whats the diference on all these

[modern cloud]

onnx is for non nvidia

fcpe is lightweight but less precise

rmvpe is the most robust and precise

crepe is usually not suggested and very sensible to noise

[mortal sluice]

so pretty much just use rmvpe

[modern cloud]

if u want the best robustness and precision, yes, rmvpe_onnx in ur case

well, do you need any other help?

[mortal sluice]

i think thats it, really now its just trial and error and finding models

thx for all the help