Your setup isn't spelled out - stereo inputs vs. a single channel. I'm guessing vocal means you have just the L/MONO A/D input plugged into a microphone -- but who knows. Adjust the details as necessary. I'll keep those details that dictate mono vs. stereo and those choices vague.

Press:
1) [PERFORMANCE] button
2) [EDIT] button
3) Touchscreen: "Audio In" -> "Mixing"
You'll see a screen like page 147 of https://jp.yamaha.com/files/download/other_assets/9/812529/montage_en_rm_a0.pdf

Set the A/D output to a USB channel of your choice (or stereo pair if you're using stereo).

In Cubase, setup the Audio inputs by pressing F4 on a Windows PC - maybe the same thing for mac. Cubase 8 calls this "VST Connections" from the "Devices" top menu.

If, within Cubase, you're using the default "Stereo In" bus - then this is everything which is coming out Main L/R (the digital/USB carbon-copy of Main L/R analog out). So if you can hear your MIDI file playing out speakers connected to Main L/R (or headphones connected to Montage) - then that is all going to spill out the default "Stereo In" bus of Cubase which is set the Main L/R.

This doesn't mean you couldn't use the default Stereo In (Main L/R). You would just have to re-route all the PARTs (which are playing instruments as dictated by your MIDI file) AWAY from Montage's Main L&R output. This would mean routing these PART(s) to Montage's (analog) Assignable Outputs perhaps. You'd lose some effects as a consequence -- that's the nature of Assignable Outs (not the same as Main outs). However - the same can be said about the digital outs as well. If you use discrete USB channels - then the system/master effects will not be applied. You'll have to keep your effects strictly in the insertion side if you want effects on your A/D routed to discrete USB channels (USB1-30).

You can still post-process by taking a recording (of your USB1-30 audio track(s)) and then play this back into Montage (digital input) and record Main L/R this second pass with Montage applying system and/or master effects.

As a general rule - you cannot simultaneously have two mixes that independently use system and master effects. There's only one output stream for the master and system effects (meaning variation and reverb). If you have two streams you want separated (like your instruments played by the MIDI file and your vocal track) - you have to make a decision which is priority for effects. If the MIDI playback is "scratch" for vocal reference - then I would route those out the Assignable L&R outputs and live with the loss of effects applied - and setup what I wanted (insertion + system + master effects) for the A/D input. A/D input routed out Main L&R so it gets the benefit of all effects - and every PART output routed to Assignable L&R outputs.

If you want to record a vocal scratch track where you do not care about the effect flexibility - OR your A/D input (vocals) do not need anything other than insertion effects - then follow the pictures above to isolate the vocals to discrete USB channels (choose 1-30) and leave your PARTs' (instruments) outputs to Main L&R. This option is easier to setup since there's only one A/D output to fool with -- whereas there may be up to 16 PART outputs to deal with if you had to move them off of Main L&R for the other option.