It depends on: the medium, the frequency, the shape of the soundwave, how it is generated (in this case: fast rotating props) the degree to which the frequencies line up (or not) which will result in beats, the phase in case that they do line up, reflections, the distance to the source(s) from the position of the receiver, the air pressure, the direction of the wind and a ton of other factors that result in more or less noise at the receiver.
To say it doesn't behave linearly would be too nice. It's not quite random, but systems quickly become so complex (2 drones: 8 rotors!) that modeling them is non-trivial. Look at the way a steady wind blowing on water can perturb it and how complex the resulting waves are. Sound is similar, you can start off with a 'clean' source (say, an electrostatic speaker) and then you place it in a room and suddenly there are all kinds of places in the room where the sound is louder or where it almost completely disappears. Then you add a second speaker, at a little distance to the first. Put them in parallel to make it simple. Now where you place your speakers relative to each other is almost as important as the distance between the two of them. If you place both speakers so far away that the sound of an individual speaker is below the noise floor of the receiver you can usually manipulate things such that the two signals crest right at the receiver. Alternatively, you can place them so close that either one of them individually is clearly above the noise floor but placing the second one strategically to cancel out the sound of the other (besides those pesky reflections) is a possibility as well.
Sound in free space is complex, and with multiple sound sources and a complex environment it quickly gets up to a level that is probably best described as 'chaotic'.