In short
When a source of sound, an observer, or both move through the medium, the frequency heard by the observer is different from the frequency emitted by the source. For motion along the line joining them, in a medium where sound travels at speed v:
Sign convention: the upper signs correspond to motion that reduces the distance between source and observer (source moves toward observer, or observer moves toward source); the lower signs to motion that increases it. Equivalently: the top sign is +v_o when the observer moves toward the source, and the bottom sign is -v_s when the source moves toward the observer.
Three qualitative facts to remember:
- Motion of the source changes the wavelength of the emitted wave.
- Motion of the observer changes how quickly they encounter fixed wavefronts.
- The medium (air, water) is the reference frame that matters for sound — wind carrying air past both source and observer changes the result too.
For electromagnetic waves (light, radio, radar) there is no medium, so only the relative velocity matters; the relativistic formula is f' = f\sqrt{(1 - \beta)/(1 + \beta)} for motion along the line, where \beta = v_{\text{rel}}/c. For v_{\text{rel}} \ll c this reduces to the classical formula with v = c.
The effect is why an ambulance siren drops pitch the instant it passes, why traffic radar can read the speed of a speeding Maruti on NH-8, why obstetric Doppler ultrasound reveals a foetal heartbeat, and why the light from distant galaxies is red-shifted — the signature that taught astronomers the universe is expanding.
Stand at the corner of a busy Mumbai road in the afternoon and wait for an ambulance. You hear it long before you see it: a high, urgent wail rushing up the street toward you. For a heartbeat, as the ambulance is directly level with you, the pitch holds steady. Then — the instant it passes — the pitch drops, abruptly and unmistakably, to a lower, slower moan as the vehicle recedes toward Bandra.
The siren did not change. Exactly the same spring-loaded horn, the same circuit, the same air flow, the same vibrating membrane producing the same frequency. What changed is the frequency you heard. While the ambulance was approaching, each successive pressure crest had less distance to travel than the one before — it was fired from a point slightly closer to you — so the crests piled up in front of the ambulance and reached your ear more often than they were being emitted. After the ambulance passed, each crest had further to travel than the one before, so they spread out and arrived at your ear less often than they were being emitted. The source was unchanged; the geometry of emission changed.
That is the Doppler effect. It is the first and simplest example of a result that threads its way through physics: a frequency that is emitted in one reference frame is not the same as the frequency received in another frame, whenever there is relative motion. It turns up in radar speed guns on National Highway 8, in the foetal heartbeat your doctor listens to through an ultrasound wand, in the red-shifted spectra of distant galaxies that told Edwin Hubble the universe is expanding, and in the GPS corrections that keep Indian Railways timetables synchronised to satellite atomic clocks. This article derives the formula from first principles and walks through each of the cases.
You should have read sound waves — nature and propagation first — we use the speed of sound v throughout — and introduction to waves for the basic relation v = f\lambda.
The simplest case: wavefronts crowd in front of a moving source
Take a loudspeaker sitting still in still air, emitting a continuous tone of frequency f. It produces a wavefront every T = 1/f seconds. Each wavefront is a spherical surface of compression expanding outward at the speed of sound v. At any instant, the successive wavefronts nest around the loudspeaker like concentric shells, spaced \lambda = v/f apart.
Now put the loudspeaker on a cart and roll it to the right at speed v_s through the still air, with the speaker still emitting f wavefronts per second. What happens? Each wavefront, once emitted, travels outward through the air at the speed of sound v — it has no memory of the speaker's motion, because the air itself is not moving. But the place where each new wavefront is born keeps shifting to the right.
The wavefront emitted at t = 0 has had one period of time to expand. At t = T, this wavefront is a sphere of radius vT = \lambda centred on the original speaker position. But by t = T, the speaker has moved to the right by v_s T. The wavefront emitted at t = T is born at this new position. So the distance between the t = 0 wavefront (whose leading edge to the right is at x = \lambda) and the t = T wavefront (born at x = v_s T, both leading rightward) is
Why: the front wave has raced out a distance vT in the time the source has slid forward by v_s T; the gap between them is the difference. In front of the moving source the wavelength is squeezed down.
Behind the source, exactly the same argument gives a stretched wavelength
because each new wavefront is born further from the previous one's expanding rear edge.
The frequency an observer hears is the frequency with which wavefronts arrive at their ear. If the air is still and the observer is stationary, a wavefront moving at speed v with spacing \lambda' arrives at the observer every \lambda'/v seconds, so the observed frequency is
Substitute the two expressions for \lambda':
Source moving toward a stationary observer:
Source moving away from a stationary observer:
In the approaching case, the denominator is less than v, so f' > f — higher pitch. In the receding case, the denominator is larger, so f' < f — lower pitch. The moment the source passes the observer, the formula switches from (1) to (2): the pitch drops abruptly from f\cdot v/(v - v_s) to f\cdot v/(v + v_s), a jump that the ear registers as a sudden downshift. That is the ambulance siren.
Moving observer instead of moving source — a subtly different calculation
Now hold the source fixed and let the observer move. The wavefronts emitted by a stationary source are perfect concentric spheres, exactly \lambda = v/f apart. Nothing about them is changed by how fast the observer is moving — the wavelength is set at emission and is a geometric fact of the air.
What changes is the rate at which the moving observer passes through successive wavefronts. Consider an observer moving toward a stationary source at speed v_o through still air. In one second, two things happen: the wavefronts sweep past the observer at v from the source's side, and the observer sweeps into new wavefronts at v_o from her own side. Relative to her, the wavefronts are rushing at her at v + v_o. The wavelength is unchanged at \lambda = v/f, so the number of wavefronts she encounters per second — the observed frequency — is
Why: frequency observed equals relative speed of wavefronts divided by wavelength. When she moves toward the source, their relative closing speed is their sum; when she moves away it is their difference.
Similarly, an observer moving away from a stationary source observes
The important subtlety: a moving source and a moving observer are not symmetric
Equations (1) and (3) look superficially similar — both predict a pitch rise when they close on each other — but they are not the same formula. Compare:
- Source approaching observer at v_s, observer still: f' = \dfrac{v}{v - v_s}f.
- Observer approaching source at v_s, source still: f' = \dfrac{v + v_s}{v}f.
If you expand both for small v_s/v, they agree to first order:
The first-order part 1 + v_s/v is the same, but the higher-order parts differ. For an ambulance doing 20 m/s through still air (about 72 km/h), v_s/v \approx 0.058, and the difference between the two formulas is about 0.3% — not audible to the untrained ear but measurable.
Why the asymmetry? Because sound propagates through a medium, and the medium picks out a preferred frame. The formula for a moving source knows where the wavefronts are born (they are generated at moving positions); the formula for a moving observer knows where the wavefronts are detected (they are encountered at moving positions). Birthing and encountering are not the same operation, so the formulas differ. If sound had no medium — like light — only the relative velocity would matter and this asymmetry would vanish. It is the presence of the medium that creates it.
That is why, strictly speaking, there are four formulas you should remember (or, equivalently, one combined formula with two \pm signs).
The combined formula — source and observer both in motion
Put the two effects together. Let the source move at v_s and the observer at v_o, both along the line joining them. Each moves through the medium. Reset the sign convention to something sane: take the positive direction as "from source to observer" at the moment you are measuring.
With that convention:
- If the source moves toward the observer, v_s > 0.
- If the observer moves toward the source (i.e., in the - direction relative to our axis), v_o > 0.
Then the observed frequency is
where the +v_o in the numerator comes from the observer-moving-toward case (equation 3) and the -v_s in the denominator from the source-moving-toward case (equation 1).
If the observer moves away from the source, flip the sign of v_o: the numerator becomes v - v_o. If the source moves away from the observer, flip the sign of v_s: the denominator becomes v + v_s. This is the sign convention that gives the textbook form
with the understanding that you pick the upper signs for approaching motion (reducing the source-observer distance) and the lower signs for receding.
A single consistent rule: at each of v_o and v_s, the sign is chosen so that motion which reduces the distance gives a larger f', and motion which increases the distance gives a smaller f'. Higher frequency when approaching, lower when receding — that is always the correct direction. If your signs make the formula point the wrong way, flip them.
Visualising the compressed and stretched wavelengths
Worked examples
Example 1: Ambulance on Western Express Highway
An ambulance moving at 20 m/s (72 km/h) sounds a siren at 660 Hz. You are standing stationary on the footpath, watching it approach. The air temperature is 25 °C. (a) What frequency do you hear as the ambulance approaches? (b) What frequency do you hear the instant after it passes? (c) What is the total "pitch drop" you hear as it passes?
Step 1. Speed of sound at 25 °C.
Why: the rule of thumb from sound waves — nature and propagation gives v to within 1% at ordinary temperatures. Precision here is limited by the assumed 25 °C anyway.
Step 2. Apply equation (1) for approach (v_o = 0, source toward you):
Step 3. Apply equation (2) for recede:
Step 4. The audible drop.
Why: the pitch does not drop smoothly — it jumps by \Delta f at the moment of passing, because the formula switches from equation (1) to (2) in an instant. Subjectively this is what the ear picks up as the characteristic "DOOOoo" of an emergency siren going past.
Result: You hear about 700 Hz approaching, 624 Hz receding — a drop of 76 Hz, roughly a musical semitone plus a quartertone (about 1.5 semitones).
What this shows: the effect is substantial even at ordinary city traffic speeds. If you had a piano app on your phone and a good ear, you could estimate the ambulance's speed from the pitch drop alone — Doppler-effect speed radar uses exactly this principle (though with radio waves and electronics instead of ears).
Example 2: Running toward a train whistle
A train at Chennai Central blows its whistle at a steady 300 Hz. A student runs toward the stationary train at 4 m/s on the empty platform. The air is at 20 °C, so v = 343 m/s. What frequency does the student hear? Now suppose the student stops and the train starts moving toward her at 4 m/s — what does she hear?
Part (a): Student moving toward stationary train. This is equation (3) with v_o = 4 m/s:
Why: the wavelength in the air is fixed by the stationary source at \lambda = v/f; the student's forward motion just means she encounters wavefronts more quickly. Nothing about the wave in the air has changed.
Part (b): Train moving toward stationary student. This is equation (1) with v_s = 4 m/s:
Step 3. Compare.
Both parts give f' \approx 303.5 Hz to four-figure precision. But expand more carefully:
Why: the two answers agree to first order in v_s/v (both give 1 + 4/343), but differ in the second-order term: the source-moving case has an extra 0.014% correction. The difference is 0.04 Hz — far below any audible threshold but a real, measurable difference.
Result: In both scenarios, the student hears approximately 303.5 Hz. The answers agree to 0.01%; the tiny difference is the signature of the medium's rest frame.
What this shows: for small speeds compared to the speed of sound, "who is moving" barely matters — the classical Doppler effect is nearly symmetric in v_s and v_o. But for high speeds (a supersonic jet), the asymmetry grows dramatically, and the two formulas diverge. For light, where there is no medium, the two cases must be identical; that demand is what forces the classical formula to break down at v \to c and be replaced by the relativistic expression (mentioned at the end of this article).
Example 3: Bat echolocation — Doppler on the return trip
A flying fox (Indian fruit bat) emits an ultrasonic chirp at 40 kHz while flying at 5 m/s toward a stationary wall. The chirp reflects off the wall and returns to the bat. What frequency does the bat hear on return? Temperature is 30 °C.
Step 1. Speed of sound.
Step 2. Frequency received by the wall (bat = moving source, wall = stationary observer).
Why: the bat is a source moving at 5 m/s toward the stationary wall. Use equation (1) with v_s = 5.
Step 3. Frequency received by the returning bat (wall = stationary source emitting f', bat = observer moving toward it at 5 m/s).
Why: the wall re-emits the reflected sound at the same frequency f' it received (a hard reflector does not absorb or shift the frequency). For the return trip, the bat is a moving observer approaching a stationary source — equation (3) with v_o = 5.
Step 4. Total frequency shift.
This is roughly twice the one-way shift — the "factor of 2" of reflection Doppler.
Step 5. Express as a fractional shift.
about 2.9%. Why: for v_{\text{bat}} \ll v, the round-trip shift is \Delta f/f \approx 2 v_{\text{bat}}/v, exactly what you'd use to estimate the bat's speed from the reflected chirp.
Result: The bat receives a reflected chirp at about 41 162 Hz — a shift of about 1.2 kHz above the emitted 40 kHz. By measuring this shift, the bat (with its specialised cochlea) can compute its approach speed to the wall — and distinguish a wall from a flying insect, which would reflect a much smaller shift (insects move slowly).
What this shows: the same principle is used in police radar on National Highway 8. A radar gun at a speed check beams a microwave signal at frequency f (typically \approx 10.5 GHz) at an oncoming Maruti. The signal reflects, and the gun measures the returned frequency f''. From \Delta f = 2 v_{\text{car}} f/c the car's speed is extracted instantly; the challan arrives by SMS. The physics is identical to the bat problem, just with v replaced by the speed of light c and the air by vacuum (and the relativistic version of the formula, which we touch on in the going-deeper section).
Common confusions
-
"Doppler shift means the source is producing a different frequency." No. The source is doing exactly what it always does. The frequency of the sound in the air ahead of the source and behind it genuinely differ (wavelengths are compressed/stretched), but the source's vibration rate is unchanged. What changes is the geometry of emission. If you put a microphone on the ambulance itself, it would read exactly the siren's emission frequency f — no Doppler shift between co-moving observer and source.
-
"Doppler shift needs the source to be accelerating." No. It needs nonzero relative velocity; acceleration has nothing to do with it. A car cruising at a constant 60 km/h produces a steady Doppler-shifted pitch relative to a stationary observer. The pitch only changes — the "drop" you hear as the car passes — because the geometry changes (approaching becomes receding). A car moving transversely past you, with zero component of velocity toward or away, produces no Doppler shift at all (at the moment of closest approach).
-
"The observer has to be moving through the medium." The observer's motion must be measured relative to the medium (still air, or still water), not relative to some arbitrary frame. If the air itself is moving — a strong monsoon wind — that changes the effective v seen by observer and source. In most school problems the air is at rest, so "moving through the medium" and "moving in any sensible frame" are the same thing; be careful when a question mentions wind.
-
"For light, the same formula works with v replaced by c." It does, to first order — but light has no medium, so the asymmetry between moving source and moving observer cannot be real for electromagnetic waves. The correct formula is relativistic: f' = f\sqrt{(1 - \beta)/(1 + \beta)} for approach (where \beta = v/c), which depends only on the relative velocity. For \beta \ll 1 this expands to f' \approx f(1 - \beta)^{1/2}(1 + \beta)^{-1/2} \approx f(1 - \beta), the classical result with v = c. Use classical Doppler for radio waves and radar whenever v/c \ll 1; the full relativistic form matters only for astronomical objects moving at appreciable fractions of c.
-
"Doppler effect applies only to sound." The effect is geometric and universal. It applies to any wave — sound, light, water waves, radio — whenever there is relative motion between source and receiver. The medium-dependence is specific to classical sound; the geometric compression of wavefronts happens in every case.
-
"If you move at the speed of sound, you hear the ambient sound at zero frequency." This is the case v_o = v in equation (4) moving away from a stationary source: f' = f(v - v)/v = 0. A pilot flying exactly at Mach 1 away from a sound source would indeed never hear that source, because the wavefronts cannot catch up. But flying toward a stationary source at v_o = v, equation (3) gives f' = 2f — you encounter twice as many wavefronts per second. No singularity on that side.
-
"A source moving at the speed of sound produces infinite pitch." Equation (1) gives f' \to \infty as v_s \to v. Physically, the wavefronts pile up on top of each other at the source's leading edge, producing a shock wave — the "sonic boom" of a supersonic jet. The classical formula breaks down at this point because the small-amplitude linear-wave approximation fails; the shock is a finite-amplitude disturbance outside the scope of equation (1). But the approach to v_s = v is real: the wavefronts do crowd asymptotically closely, which is what creates the sonic boom.
Applications — where the Doppler effect shows up in India
Traffic radar on NH-8 and beyond. Indian state highway police use Doppler radar guns at stretches of NH-8 (Mumbai-Ahmedabad) and NH-44 (Srinagar-Kanyakumari) to enforce speed limits. The gun emits a microwave pulse at roughly 10.525 GHz (X-band) or 24.125 GHz (K-band), and measures the Doppler-shifted return. A fractional shift of 2v_{\text{car}}/c reveals the car's speed; electronics inside the gun convert this to km/h on the display. A Swift Dzire doing 100 km/h = 27.8 m/s produces a round-trip shift of 2 \times 27.8 \times 10.525 \times 10^9 / (3 \times 10^8) \approx 1950 Hz — well within the resolution of the gun.
Doppler ultrasound in medicine. AIIMS, PGI Chandigarh, and every major teaching hospital in India uses Doppler ultrasound to image blood flow. A transducer at 2–10 MHz beams ultrasonic pulses into tissue; red blood cells moving through arteries reflect the pulses with a Doppler shift proportional to their velocity. The shift is typically a few hundred to a few thousand Hz — in the audible range — and is often converted to an audible whooshing sound that obstetricians listen to when checking a foetal heartbeat. The mathematical formula is equation (6) applied twice (once for the moving blood cell as reflector).
Doppler weather radar. The India Meteorological Department operates S-band Doppler weather radars at stations across India, from Sriharikota (for ISRO launches) to Mumbai to Chennai to Guwahati. These radars beam microwaves at around 2.8 GHz into the atmosphere; raindrops reflect the signal, and the radial velocity of the raindrops — computed from the Doppler shift — reveals the wind field within storm clouds. A velocity pattern indicating rotation reveals a developing cyclone.
Red shift and the expanding universe. Light from a distant galaxy is observed at a slightly longer wavelength — lower frequency — than it was emitted at, and the further away the galaxy the larger the shift. For a galaxy receding from us at speed v much less than c, the classical Doppler formula gives \Delta\lambda/\lambda \approx v/c. Edwin Hubble's 1929 discovery that v scales with distance — v = H_0 d, where H_0 is Hubble's constant — is the founding observation of modern cosmology. Meghnad Saha and others in India contributed early to the spectroscopic techniques that make red-shift measurement possible. The red shift is Doppler plus an additional cosmological expansion effect; at small distances the two agree.
GPS and IRNSS timing corrections. India's own satellite navigation system, NavIC (IRNSS), relies on atomic clocks in satellites. As the satellite passes overhead, the Doppler shift of its carrier frequency (as received on your phone) encodes how fast it is approaching or receding from you. The receiver combines this with several other satellites' Doppler data to pinpoint your position. The correction for relativistic Doppler effects — both special and general relativistic — is essential for sub-metre accuracy.
If you came here to understand the Doppler effect and its formula, you have what you need. What follows is the derivation of the full formula including wind, the oblique case where the motion is not along the source-observer line, the relativistic Doppler effect for light, and the sonic-boom limit where the classical formula breaks down.
Full classical formula including wind
If the medium (air) is itself moving at velocity w relative to the ground, in the same direction as the source-observer line, the source's motion relative to the medium is v_s - w and the observer's is v_o - w. Substituting into (6):
A tailwind (w > 0) effectively slows the apparent speed of sound from the source's viewpoint and speeds it up from the observer's — the two effects don't simply cancel. That is why voices carry much further with the wind than against it (a common observation near the Coorg coffee estates at dawn).
The oblique case — motion at an angle
If the source is moving at speed v_s along a line at angle \theta to the line joining source and observer, only the component along the line-of-sight contributes to Doppler shift. The generalisation of equation (1) is
At \theta = 90° (transverse motion), \cos\theta = 0 and classical Doppler predicts zero shift. This is the moment of closest approach of a passing ambulance — right as it is level with you, the pitch holds steady, because its velocity is momentarily perpendicular to the line to your ear.
For light, however, there is a transverse Doppler effect — a purely relativistic one — with f' = f/\gamma where \gamma = 1/\sqrt{1 - \beta^2} is the Lorentz factor. For \beta \ll 1 this is a second-order effect, f' \approx f(1 - \beta^2/2), and was confirmed in the Ives-Stilwell experiment (1938).
Relativistic Doppler for electromagnetic waves
Light has no medium, so the Doppler formula must depend only on the relative velocity u between source and observer. Applying special relativity to the wavefronts gives, for motion along the line joining them at approach speed u:
Observer and source are now fully symmetric — as they must be for light — and the classical asymmetry of equations (1) and (3) is no longer present. Expand for small \beta = u/c:
which recovers the classical result with v = c. Higher-order corrections become important for astronomical objects moving at \beta \gtrsim 0.1.
This formula explains why the spectra of distant quasars are heavily red-shifted (some at z = \Delta\lambda/\lambda > 4, meaning \beta > 0.94) and how exoplanet-hunters detect the tiny reflex motion of a star tugged by an orbiting planet — a few m/s of wobble in a star at hundreds of light-years away, extracted by measuring sub-milli-ångström Doppler shifts in its absorption spectrum.
Why classical Doppler breaks at v_s = v — and what replaces it
As v_s \to v in equation (1), f' \to \infty. Physically, wavefronts pile up at the source's leading edge; the assumption of small-amplitude linear waves — which was baked into every step of the derivation — fails. What actually happens is that the wavefronts coalesce into a single cone-shaped shock wave — the Mach cone — whose half-angle \alpha is given by \sin\alpha = v/v_s. For v_s > v, the source travels faster than its own sound can escape; observers outside the Mach cone hear nothing until the shock passes, then a single loud "crack" — the sonic boom. Supersonic jets and even some artillery shells produce this. The famous "Concorde boom" heard along the Bay of Bengal in the 1980s was of this kind.
The mathematics of shocks is nonlinear and outside the scope of the classical Doppler derivation. But the point at which the formula breaks — v_s = v, Mach 1 — is exactly the regime where the linear approximation fails. Every time a formula diverges, it is signalling that new physics is required.
Where this leads next
- Sound Waves — Nature and Propagation — the speed of sound v used throughout, and the medium-dependence that underlies the classical Doppler asymmetry.
- Vibrations of Strings and Pipes — the sources of most musical Doppler experiments: a violin passing on a train, a flute moving past a recorder.
- Intensity and Loudness of Sound — why the amplitude of a moving source's sound also changes with distance (inverse-square law), and how that combines with frequency shift to shape what you perceive.
- Principle of Superposition — the Doppler effect combined with reflection produces complicated interference patterns around a moving source, a key effect in radar and sonar.
- Reflection and Transmission of Waves — reflection-Doppler, the effect used in police radar and ultrasound imaging, described briefly in Example 3 and rigorously in that article.