I do the following for recording in OBS Studio with multi track audio. I also use it for streaming too.
Track 1 = Game/Desktop audio
Track 2 = Mumble/Teamspeak/Discord
Track 3 = My Microphone
Track 4 = The three above all in one track
Makes it nice when I edit. Because I can adjust or mute individual audio tracks if I want. The stream gets track 4 only so they hear everything already mixed.
I keep my game/desktop on track 1 because my video editor doesn't support multiple tracks and just uses the first track. I extract the audio for tracks 2 and 3 with a GUI shell I wrote that uses ffmppeg to do the extract.
I have track 4 with all the first three tracks mixed together so I can listen with VLC to the capture and select that track and hear everything. Nice for checking the recording, looking for moments I want, or just jumping around quickly. VLC doesn't play multiple audio tracks at one time, so having it all mixed on track 4 allows me to listen to everything.
In order to do the multi track audio with more than just your game on one track and microphone on another you will need either Voicemeter or Virtual Audio Cable
http://vb-audio.pagesperso-orange.fr/Voicemeeter/index.htm
http://software.muzychenko.net/eng/vac.htm
I use the former one. Use to use VAC, but have switched to Voicemeeter/Bananna
You basically create virtual audio devices with these. You then can direct where the output goes.
So I have a virtual audio cable that is my windows default sound device. So the sound of that is games and everything. I have that being output to my Sound card and my headphones.
I have another virtual cable audio device that my mumble/discord/teamspeak output to. That device is directed to my headphones sound card but does not go to the default audio device (which would get to the sound card/headphones but would be picked up in the default audio channel).
I then have OBS record default channel (the first virtual audio device) on track 1
I have it record the second virtual audio device on track 2 (mumble/discord/teamspeak)
And my microphone on track 3.
You then just go into teamspeak/mumble/discord and specify that second virtual audio device as your output device.
Hope that makes sense.