You don’t need to loopback like that, you can just create a dummy output in Cubase (Audio connections):
In that example i’ve created one call ‘Capture’, you don’t have to set any hardware outputs for it to exist.
Then if you see the other yellow highlights, I set the original vocal track (With variaudio applied) to output to the ‘Capture’ output channel.
Then on a new audio track (RT Render), set ‘Capture’ as it’s input. So it will be able to receive any audio routed to ‘Capture’.
You won’t hear the original vocal track until you record monitor the receiving track, and can simply record in realtime. You’ll get much better quality and not have to balance levels doing it this way.
I use this method for VST instruments and recording them as though they are outboard hardware - as I much prefer to deal with realtime performances in audio form as it removes any temptations for me to start spending time correcting MIDI notes, and relying on quantisations.