@larsmb the examples I remember were based on the concept the car had a siren of a fixed frequency.
I think trying to compute the relative frequency shift of what is likely white/pink noise, would extremely difficult.
Especially if the car was accelerating, then the dominant noise would be increasing in frequency anyway.
Any form of radar would be far easier and more reliable.
@intrbiz No, because you could still compute the relative frequency shift of the sound as it passes you by.
(My physics teacher had us do the math - the person heard a quint, I recall. Obviously real audio processing is harder than a physics exam, but it should be possible.)