Well, we’re getting there…
So as I understand it, your sound modules’ audio outputs are connected to an external mixer correct? And then the main stereo out of that external mixer is then connected to inputs on your audiobox?
And then your monitor speakers are connected to outputs on your audiobox, correct?
Finally, I assume you’re using an Audiobox USB?
The first thing I would say, which probably has little to do with your problem, is that if you ARE using the AudioBox USB, it is not designed to accept line level inputs. Are you somehow bringing the line level outputs from your mixer down to either mic or instrument level with a direct box BEFORE you go into your AudioBox? Setting the gain level properly for a line level signal with the AudioBox would be very persnickety. See this knowledge base article from Presonus:
The second thing is… are you paying attention to where the “Mix” knob is set on the AudioBox? It’s POSSIBLE that you are doubling up the audio when sending MIDI to your sound modules. If the “Mix” knob is set to anything BETWEEN “Inputs” and “Playback” you might be sending audio both directly through the audiobox to the outputs, AND audio from Cubase through the outputs. If you are monitoring through Cubase, this would have the effect of doubling the volume (and probably chorusing / delaying) of the sound from the modules that you are sending MIDI to. And then after you’ve recorded the audio (bouncing down from MIDI) when Cubase plays it back you are only hearing what you SHOULD have heard in the first place, which is the module’s pure playback without the doubling, and at the proper volume.
My guess is that it’s this, or something like this, that’s happening. The REAL sound is what you hear at the reduced volume. The problem is likely in the monitoring before and during recording, where you are hearing more than you should. This could be happening somewhere else in the chain, depending on how you’ve set it up. If you’re monitor speakers (or amp) is connected to another intervening mixer, for example.
Again, I’m just guessing… there are a lot of potential variables.