Have an independent level of the signal for both audio sources inside VM: voice and sound fx/meme. Have a separate slider for background level. Inside every sound, have a sort of Post Fader Aux Sends that will control independently the amount of signal level sent to 2 outputs and a master fader per output with mute and picometer:

In a second step, include third parties sound buses, to become the last step before connecting to OBS/streaming service.