The air in the car, moving with the car, and at rest with the car , of course, has an interface with the air outside the car’s ONE , slightly-open window moving at the car’s velocity.
If there is no wind and the car is relatively aerodynamically ‘smooth’ that interface is like a smooth drum face which is not vibrating , i.e., not being ‘struck’ by anything…. and their shall be no [or very little] ‘sound’. If there is some wind , but it is moving in line with the car(towards or away from the moving car)... there also would be minimal ‘sound’ created by ‘vibration’ of that interface-drum-membrane.
If there is some wind moving NOT in line with the motion of the car AND especially if the wind is , as one would expect , slightly varying in its speed, THEN one gets turbulent interference at that ‘membrane’ of air… and ‘vibration’ of that interface and SOUND… varying in pitch with the slight change in wind speed AND change in direction of the car with respect to that wind.
With BOTH windows open…. you have turbulent flow inside the car causing the sound, especially at the edges of the open windows….....