An ambulance screams towards you, siren wailing at a steady, urgent pitch. The instant it flashes past and starts pulling away, the note drops — the same siren, still blaring at exactly the same frequency, suddenly sounds lower. You have heard it a thousand times: the falling neeeeeoowww of a passing siren, a racing car, a train whistle. The driver, riding along with the siren, hears no change at all. What you hear depends on how the sound source is moving relative to you. That shift in pitch is the Doppler effect, and it is one of the most useful facts in all of physics — it is how radar guns clock speeding cars, how astronomers weigh the expansion of the universe, and how an ultrasound machine watches blood flow through your heart.

The whole idea rests on one picture, worth holding onto: sound travels through the air at a fixed speed, no matter how the source is moving. The source cannot make its sound go faster by chasing after it — it can only crowd its waves closer together in front and string them out behind. Below we will draw that picture, turn it into a formula, and then confront the subtlety that catches almost everyone: moving the source and moving the listener are not the same thing, even when the closing speed is identical.

The physical picture: wavefronts on the move

A sound source sends out a fresh crest (a wavefront) at regular intervals — one every period T = 1/f. Each crest, once released, spreads outward as a growing sphere at the speed of sound v, centred on the spot where the source was at the moment it let that crest go. If the source sits still, all those spheres share one centre — evenly spaced rings, the same wavelength in every direction, and everyone around hears the true frequency f.

Now let the source move. Between releasing one crest and the next, the source has crept forward. So each new circle is drawn from a point a little further along than the last. Ahead of the source the circles bunch up — their leading edges crowd together — and behind it they fan apart. The speed of each wave is unchanged (the air does not care that the source is moving); it is the spacing, the wavelength, that is squeezed in front and stretched behind. Shorter wavelength at the same speed means a higher frequency heard ahead; longer wavelength means a lower frequency heard behind. Reveal the figure step by step to watch the crests pile up in the direction of travel.

This is why a passing siren does not slide smoothly from high to low. As the ambulance approaches, every crest is compressed by the same factor, so you hear a steady high note. As it recedes, every crest is stretched by the same factor, so you hear a steady low note. The famous swoop happens only in the brief moment it sweeps past, when the approach turns into a recession.

The general formula, and its signs

Call the speed of sound in air v (about 340\ \text{m/s}), the observer's speed v_o, and the source's speed v_s. The frequency the observer actually hears, f', is related to the frequency the source emits, f, by

The pile of \pm and \mp signs looks fearsome, but there is a single rule that fixes every one of them: motion that brings source and observer together raises the pitch; motion that carries them apart lowers it. You never have to memorise which sign is which — you just ask "does this motion close the gap or open it?" and pick the sign that pushes f' the right way.

The two effects can act at once — you can be driving towards an ambulance that is also driving towards you — and then both the top and bottom signs are set to "approach", stacking a double shift. But notice something important about the shape of the formula: the observer's speed v_o sits in the numerator, while the source's speed v_s sits in the denominator. Those are not the same operation, and that is the heart of the next section.

Why moving the source ≠ moving the observer

Here is the surprise. Suppose the closing speed is u either way. If the source rushes towards a still observer at u, the observer hears

But if instead the observer rushes towards a still source at the same u, they hear

These are different numbers. Only for small speeds (u \ll v) do they nearly agree — both give roughly f(1 + u/v). The reason is physical, not a quirk of algebra. When the source moves, it changes the wavelength of the sound waiting in the air — it literally lays down crests closer together — and the wave then arrives at the still listener at the ordinary speed v. When the observer moves, the wavelength already sitting in the air is untouched; instead the listener runs into the crests faster, changing the rate at which they sweep past. Different mechanisms, different formulas.

There is a deep reason this asymmetry exists at all: sound has a medium — the air — and that air defines a special, preferred frame of rest. "Who is moving with respect to the air" genuinely matters. This is exactly the point where sound differs from light: the relativistic Doppler effect for light has no medium and no preferred frame, so it depends only on the relative velocity of source and observer — moving one is identical to moving the other. Sound is not so tidy, and this page is all about the tidier-looking but subtler classical case.

The graph makes the split vivid: both curves show f'/f as the moving thing speeds up, but the source-moving curve bends upward ever more steeply (its denominator is shrinking towards zero — a runaway as v_s \to v), while the observer-moving curve is a plain straight line.

Worked examples

Example 1 — the ambulance approaching. A siren emits f = 700\ \text{Hz}. The ambulance drives towards a stationary you at v_s = 30\ \text{m/s} (about 108\ \text{km/h}). The source moves towards you, so use the smaller denominator:

The note is pushed up by nearly 68\ \text{Hz} — clearly sharper.

Example 2 — the same ambulance receding. The instant it passes and drives away at the same 30\ \text{m/s}, the source now recedes, so use the larger denominator:

The pitch drops from \approx 768\ \text{Hz} to \approx 643\ \text{Hz} as it goes by — a fall of about 125\ \text{Hz}, roughly three semitones. That is the neeeeeoowww.

Example 3 — you move instead. Now the ambulance is parked with its siren sounding 700\ \text{Hz}, and you drive towards it at v_o = 30\ \text{m/s}. The observer moves towards the source, so use the larger numerator:

Compare with Example 1: the same 30\ \text{m/s} of approach gives 767.7\ \text{Hz} when the source moves but only 761.8\ \text{Hz} when the observer moves. The two really are different — moving the source shifts it a touch more — exactly the asymmetry from the last section, made numerical.

Example 4 — both at once. The ambulance approaches at v_s = 30\ \text{m/s} while you drive to meet it at v_o = 20\ \text{m/s}. Both motions close the gap, so take the larger numerator and the smaller denominator together:

The two shifts stack, pushing the pitch up by well over a hundred hertz.