As a computer scientist, I know that it is possible to decompose two signal streams (such as audio) using the SVD (Singular Value Decomposition), to get the most natural bases for the signals.
This is very useful because you could, in theory, record with two microphones to remove background noise. For example, one microphone near the speaker, and another far from the speaker. The one that is near the speaker will have a stronger speech and weaker noise, on the other hand, the far microphone will record stronger noise and weaker speech.
The problem is that I don't know of tools to accomplish this. Also I don't have a clue of what to look for, since I am not a professional in video editing. All I can find is tools to remove noise based on one audio only.