Practice with mock tests, learn from structured notes, and get your questions answered by a global forensic community, all in one place.
Forensic gunshot audio analysis deconstructs the muzzle blast, shockwave, and impact signatures of a discharge to establish the number of shots, their sequence, the weapon type, and the shooter's location using acoustic physics and time-difference-of-arrival techniques.
Last updated:
When a firearm is discharged, it does not produce a single sound: it produces a cascade of overlapping acoustic events that carry, to a trained analyst, detailed information about the weapon, the projectile, the shooter's position, and the order of events. The short crack or boom that a bystander hears is actually three physical phenomena arriving at nearly the same time. Pulling them apart acoustically and interpreting what each component means is the core of gunshot audio analysis, a field that has moved from courtroom curiosity to operational use in cities worldwide through sensor networks such as ShotSpotter.
The same physical principles that describe gunshot acoustics also underpin the broader discipline of event audio analysis: using recorded sound to reconstruct the geometry, timing, and nature of dynamic events. A vehicle collision, an explosion, a building collapse, and a gunfight all leave acoustic traces that follow the same rules of propagation, reflection, and arrival-time geometry. The analyst's job is to extract the physics from the recording before interpreting what it means.
This topic covers the three components of a gunshot signature, time-difference-of-arrival localisation methods, the ShotSpotter architecture as a real-world implementation, how suppressed and unsuppressed fire differ acoustically, the main confounders that can fool both sensors and analysts, and the landmark Dallas 2016 acoustic reconstruction as a worked example of how event sequence analysis operates under pressure in a real case.
One trigger pull, three distinct pressure events arriving at every microphone.
Understanding why a gunshot has multiple acoustic components requires thinking about what happens in the barrel and around it in the milliseconds after firing. As the propellant burns, pressure builds behind the bullet and accelerates it down the bore. When the bullet exits, the still-high-pressure gas follows it and expands explosively into the air, creating the muzzle blast: a broadband pressure impulse that radiates in all directions from the muzzle.
If the projectile is supersonic (most rifle ammunition and many handgun loads travel at 370-1000 m/s, well above the speed of sound at ~340 m/s in air at 20°C), it generates a continuous ballistic shockwave as it travels downrange. This shockwave is not a single event at the muzzle: it is a Mach cone that trails the bullet for as long as it remains supersonic. A receiver to the side of the bullet's path hears the shockwave as a sharp crack that arrives from a direction perpendicular to the flight path, not from the shooter's position. This angular difference between muzzle blast direction and shockwave arrival direction is extremely useful for determining where the shooter was standing relative to the sensor.
The impact transient is the third component. When the bullet strikes a surface it transfers energy suddenly, creating an impulsive sound whose character reflects the target material. A strike on concrete sounds very different from one on timber or water. In multi-shot incidents the sequence of impacts can help establish which shot hit what target and in what order.
The same physics that guides radio positioning works for sound, only slower.
If a gunshot is detected by two microphones at known positions, and the arrival time at each is measured accurately, the difference in arrival time constrains the source to a hyperbola in two-dimensional space (or a hyperboloid in three dimensions). In free-field conditions the constraint is given by: c × (t2 - t1) = difference in distances from source to each sensor, where c is the speed of sound and t1, t2 are the arrival times. The speed of sound depends on temperature and is approximately 340 m/s at 20°C, 333 m/s at 5°C.
A single TDOA measurement from two sensors produces one hyperbola. Adding a third sensor produces two more TDOAs (sensors 1-3 and 2-3), each constraining the source to another hyperbola. In an ideal noise-free environment, the three hyperbolas intersect at a single point: the source location. In practice, multipath reflections off buildings, temperature gradients, and cross-correlation peak estimation errors spread the intersection into a probability region rather than a point.
Putting TDOA into city infrastructure, from research paper to 911 dispatch.
ShotSpotter, founded in 1996 and widely deployed in US cities from the mid-2000s, operationalises TDOA localisation at city scale. The system uses microphones mounted on buildings, utility poles, and streetlights, typically spaced 200-300 metres apart in covered areas. Each sensor has a GPS-synchronised clock so arrival times across the network are comparable. When an impulsive event triggers multiple sensors, the company's servers run TDOA algorithms to estimate the source location and classify the event.
Location accuracy reported in the published literature varies from around 20 metres median error in open areas to over 40 metres median error in dense urban canyons where reflections complicate the TDOA estimates. The system also reports the number of rounds fired and, where waveform analysis permits, a weapon-type estimate (handgun vs. rifle, suppressed vs. unsuppressed). Alerts are typically transmitted to dispatch within 30-60 seconds of the event.
The main documented limitation is false positives from confounders. Large-scale studies have found that fireworks and vehicle backfire are the most frequent causes of false alerts. ShotSpotter addresses this with a human review step in which trained acousticians audit algorithmically flagged events before alerts are dispatched, reducing but not eliminating false positives. The proportion of gunshot incidents that go undetected (false negatives) is harder to measure, as there is no complete ground truth of all gunshots fired in a covered area.
A suppressor changes the blast, not the physics of a supersonic bullet.
A sound suppressor (commonly but inaccurately called a silencer) works by providing an expansion chamber through which the propellant gases cool and slow before reaching the air. This attenuates the muzzle blast, typically by 20-35 dB depending on the suppressor design and calibre. What it does not do is affect the ballistic shockwave of a supersonic projectile: the shockwave is generated by the bullet's velocity, not by the gases, and no suppressor changes the bullet's speed.
| Characteristic | Unsuppressed | Suppressed (supersonic) | Suppressed (subsonic) |
|---|---|---|---|
| Muzzle blast level | High (130-170 dB SPL at 1 m) | Reduced (95-130 dB SPL at 1 m) | Reduced (~80-110 dB SPL at 1 m) |
| Ballistic shockwave | Present if supersonic ammo | Present (unchanged) | Absent (bullet below Mach 1) |
| Distinctive acoustic signature | Sharp crack + boom | Crack (shockwave) + attenuated pop | Low 'thwack' or mechanical click |
| Detectability by sensor network | High | Reduced but detectable if near sensors | May not trigger detection thresholds |
Subsonic suppressed fire, using ammunition designed to keep the bullet below the speed of sound (typically below about 330 m/s), eliminates both the muzzle blast and the shockwave to an extent that can make the sound indistinguishable from non-firearm confounders at distance. Distinguishing subsonic suppressed fire from a nail gun, a pneumatic tool, or a car door slam may require physical evidence or proximity to sensors.
Reconstructing what happened and in what order, from one recording or many.
Vehicle backfire, fireworks, industrial air-powered tools, and tyre bursts share the impulsive spectral content of gunshots. Analysts use several discriminants to separate them. A genuine gunshot from a supersonic round has a double-arrival structure (shockwave then muzzle blast or vice versa depending on geometry). Fireworks tend to have a longer low-frequency tail and often arrive in cadenced bursts. Vehicle backfire typically shows a different spectral envelope with more low-frequency energy and no shockwave. These are probabilistic separations, not absolute rules.
Event sequence reconstruction, determining which shot was fired first, which microphone picked up which event, and how many distinct weapons were involved, requires cross-correlating arrival times across multiple recordings or sensors and building a timeline consistent with the speed of sound and the spatial geometry. A recording from a bystander's phone and a CCTV camera 50 metres apart will each have slightly different timing for the same events, and the difference can be used to constrain the geometry.
From which direction does the ballistic shockwave of a supersonic bullet appear to arrive at a sensor beside the bullet's flight path?
Test yourself on Forensic Audio, Video and Image Analysis with free, timed mocks.
Practice Forensic Audio, Video and Image Analysis questionsSpotted an error in this page? Report a correction or read our editorial standards.