Acoustic Drone Detection: Microphone Arrays in Practice

Acoustic detection is the humblest counter-UAS sensor and, in the right spot, one of the most useful: it's cheap, entirely passive, and it hears the one thing that defeats RF detection — a drone flying in radio silence. This guide covers how listening for propellers actually works, and why range is both its limitation and, sometimes, exactly enough.

The signature: why drones sound like drones

A multirotor's noise is dominated by its propellers — the blade-passing frequency and its harmonics, sitting on the broadband whoosh of the motors (the physics is in the propeller guide). That produces a characteristic spectral pattern — a set of tones at predictable spacings — that's quite unlike traffic, wind or birds. Detection software learns these patterns; the tones even shift with throttle, which can hint at what the drone is doing.

From sound to direction: the microphone array

One microphone can detect; it can't locate. An array of several microphones, centimetres to metres apart, can — because sound arrives at each mic at a slightly different time. Measuring those time differences of arrival (TDOA) yields the bearing to the source, and multiple arrays (or a large one) can cross bearings into a rough position and altitude. It's the same triangulation idea as RF direction finding, done with pressure waves travelling at ~343 m/s instead of radio at the speed of light — which, helpfully, makes the timing differences large and measurable with cheap hardware.

Classification: telling a drone from a lawnmower

The hard part isn't hearing a sound — it's deciding whether it's a drone. Modern systems feed array audio into a machine-learning classifier trained on drone signatures versus everything else (the same edge-AI toolkit used onboard drones, pointed the other way). Good training data is the whole game: a classifier that has never heard a particular drone, or a particular wind condition, will miss or false-alarm. Libraries like librosa handle the feature extraction; the dataset is what you actually invest in.

The real limits: range and noise

• Range. Sound intensity falls with the square of distance and is absorbed by air, so practical detection is ~300–500 m and shrinks fast for quiet drones. This is a close-in sensor, full stop.
• Wind and weather. Wind noise across the microphones is the classic killer; rain and temperature gradients bend and muffle sound. Good windscreens and array placement matter enormously.
• Urban din. Traffic, HVAC and crowds raise the noise floor and hide the signature — the same problem RF has in cities, in a different medium.
• Bigger/faster drones climb out of range quickly; acoustic favours the low, slow, close threat.

Fitting it into the system

Acoustic pairs naturally with optical: the array's bearing cues a camera to the right patch of sky for confirmation and identification, covering acoustic's inability to say which drone. Combined through sensor fusion with RF and radar, its narrow but RF-proof coverage plugs a specific, important hole. See the detection systems overview for the full picture.

Frequently asked questions

How does acoustic drone detection work?

A microphone array picks up the distinctive whine of a drone's propellers and motors. Software classifies the sound against known drone signatures, and the tiny differences in arrival time between microphones (TDOA) give the direction — and, with enough spacing, an approximate position.

How far can acoustic sensors detect a drone?

Realistically 300–500 metres in good conditions, less in wind or noisy environments. Sound falls off quickly and is easily masked, so acoustic detection is a short-range, low-altitude layer rather than a wide-area sensor.

Why use acoustic detection if the range is short?

Because it is passive, cheap, and hears autonomous drones that emit no radio signal — RF detection's biggest blind spot. Around sensitive close-in points like prison walls or a VIP venue, a short-range sensor that catches the silent drone is exactly what's needed.