#
Realism
Last update: May 3, 2025
#
Introduction
In this guide it will be explained how to increase the realism of a RVC model while using W-Okada.
We will go over installing and setting up Voicemeeter, adding VST's using Light Host and how to make your voice sound more convincing.
#
Good Practices
Using a model that is similar to your voice helps a bit. Best case scenario you have the pitch set to 0 and you don't have to act like the voice.
Index can help a bit by adding some of the models accent into the converted voice.
Acting like the voice you are trying to be also helps.
Some users think that models trained on the f0 extractors crepe and fcpe give models a "human softness" to them while rmvpe can sometimes give a "metalic sound" to them. It would be good to try models trained on rmvpe or fcpe to see which one you think sounds more real.
#
Installing Voicemeeter
Download Voicemeeter from here. You can either install Voicemeeter Banana or Potato.
Unzip the folder then run
voicemeeterprosetup.exe.Once you run it you should get this popup. Click install.
It's already installed for me so that is why it says "Remove".
- Restart your computer once you have finished downloading it.
#
Voicemeeter Setup
- Open Voicemeeter and in the top right "Hardware Out" should be flashing red.
- Click A1 and set it to your headphones.
- Then go to "Stereo Input 1"
- Click it and select 'Line 1'. If you want you can also right click "Stereo Input 1" to rename it to something.
- Once you have done that set it to output to B1 and have it on mono.
- Then go here and set both the outputs to B2
- Once you have completed all of the above steps you can now go into anything and set the mic input to "Voicemeeter Out B2".
- (Optional) If you want to decrees delay press
ctrl + ,and a menu should open. In the menu go down to where the buffer settings are and set the samples to 480. If you hear crackling or audio bugs increase the buffer size and if there is no crackling and you want less delay you can lower the buffer size more.
#
Installing Light Host
Download Light Host.
Unzip the folder then run
Light Host.exe
#
Light Host Setup
- Open Light Host and go into 'Preferences'
- Set your 'Audio device type' to "DirectSound"
- Set your output to "Voicemeeter Input" And your input to "Line 1"
- In advanced settings set your "Sample rate" to 44.1k and the "Audio buffer size" to 480. If you hear crackling increase the buffer size and if there is no crackling and you want less delay you can lower the buffer size more.
#
VSTs & Background Noise
#
VSTs
First you will need some VST's, you can find plenty of free ones at Kilohearts website. I suggest their Bitcrush, Filter and Phase Distortion but you can use whatever VSTs you want.
There is no "prefect setting" for these VSTs so set their settings to whatever you think sounds good.
The more VSTs you add the more delay there is going to be.
- To add a VST to Light Host open it then click "Edit Plugins".
- Go down to options, click it and then click scan for VST or VST3 plugins.
- It should compile a list of all your plugins. At that point you can close that menu, reopen Light Host and there should be a "Available Plugins" section. Just click plugin and it should open a sub-menu where you can add plugins.
- To edit them click on the plugin then click 'Edit'. It will open the plugin's menu so you can edit it.
#
Background Noise
- First you are going to need noise to play, you can get those at freesound or just record your own room noise using Audacity.
When you have the noise you want you are going to need to open "Windows Media Player" then drag the audio you downloaded into it.
Then you are going to need to go into windows "Sound Settings" then "Advanced sound options".
- Scroll until you see "Windows Media Player" and set its output to "Voicemeeter AUX Input".
- Then go back in "Windows Media Player" and hit play on your noise. Don't forget to turn on repeat so it doesn't randomly stop.