Last updated: May 27, 2026
Key Insights
- Latvia’s €50M acoustic drone detection network detected the Rēzekne drone but the kill chain failed — the gap was response time, not sensor performance
- Acoustic detection is the only passive modality that catches RF-silent and fiber-optic controlled drones, but physics limit range and classification speed
- European buyers should demand SAPIENT-compatible multi-sensor fusion, edge AI under 100ms latency, and a quantified answer to wind degradation — adjectives are not spec sheets
On May 7, 2026, two drones crossed into Latvian airspace from the direction of Russia, hit an empty fuel depot in Rēzekne, and brought down the country’s defence minister within days. Latvia had spent the previous year becoming the first NATO member to install acoustic drone detection along its entire eastern border. The deployment covered 87% of the 455 kilometer frontier with Russia and Belarus by September 2025. And yet a Ukrainian Shahed-type drone still got through.
If you sell or buy counter-UAS systems, this is the case study that matters right now. Here is what acoustic drone detection actually does, where it breaks down, and what the Latvian experience tells the rest of Europe about building a drone wall that works.
What is acoustic drone detection?
Acoustic drone detection is a passive counter-UAS technology that identifies unmanned aerial vehicles by the sound their propellers and motors emit. A network of ground-based microphone arrays listens for the harmonic signature of rotor blades, typically a fundamental Blade Passing Frequency between 150 and 400 Hz with harmonics extending up to 16 kHz. Machine learning algorithms classify the signal in real time and triangulate the drone’s bearing across multiple sensor nodes.
Unlike radar or radio frequency detectors, acoustic sensors are passive. They emit nothing. They cannot be jammed by electronic warfare. They detect autonomous and fiber-optic-controlled drones that radio frequency systems are blind to. And they cost a fraction of any other modality, with Ukrainian Sky Fortress nodes built for between $400 and $1,000 per unit.
Why Latvia bet on acoustic sensors
The Latvian Ministry of Defence picked acoustic detection for one reason. Shahed-type drones are radar-transparent. They are built from cheap composite materials. They fly at 50 to 100 meters altitude. They have a radar cross-section indistinguishable from a large bird.
Maj. Modris Kairišs, head of Latvia’s Autonomous Systems Competence Center, explained the tactical math in a May 26 press briefing: covering a 400 kilometer border with tactical radars would require one station every 10 to 20 kilometers, plus all the supporting infrastructure. Acoustic sensors cover the gap between radars at a tiny fraction of the cost.
The Latvian approach mirrors what Ukraine learned the hard way. Ukraine’s Sky Fortress network of roughly 14,000 acoustic sensors and the smaller Zvook system together helped achieve a 95% interception rate against Russian drone attacks. The entire Ukrainian national network cost less than $5 million to deploy. That is less than two Patriot missiles.
So what went wrong at Rēzekne?
This is where you have to separate political narrative from operational reality. Two stories ran in parallel.
The opposition framing, picked up by outlets like Pietiek and quoted in Russian media, was blunt: the acoustic systems did not detect the incoming drone, the cell broadcast alerts reached residents about an hour after the strike, and the state defence function failed.
The Latvian Ministry of Defence framing, repeated by Defence Minister Andris Sprūds and again by Maj. Kairišs in the May 26 briefing, was different: the acoustic detection system “functioned effectively” and is “quite effective” when combined with radar and optical sensors. The gap was not detection. It was the kill chain. Brig. Gen. Egils Leščinskis put it directly: the acoustic sensors picked up signals at three points that morning, but the system cannot instantly distinguish a drone from a moped or a motorcycle. By the time the trajectory confirmed a drone, it had already crashed.
Both versions point to the same operational truth, which is the lesson worth taking away.
The four physics constraints every acoustic detection buyer should understand
Anyone selling you a counter-UAS system that ignores these is selling you marketing.
- Range scales with target loudness, not with sensor quality. A quiet quadcopter at 200 to 300 meters. A loud fixed-wing Shahed at 5 to 7 kilometers. The same hardware delivers wildly different ranges depending on what it is listening for.
- Wind above 5 m/s degrades performance severely. Upwind detection can drop by 20 dB or more. Even high-quality windshields recover only 2 to 3 dB. There is no algorithmic fix for a 30 km/h headwind.
- Classification is harder than detection. Picking up a sound is easy. Calling it a drone before it crashes is the actual problem. This is what the Rēzekne incident exposed.
- Quiet drones are coming. Israel’s Aerosol G2 emits 14.9 dB at one kilometer, below rural night-time ambient noise. Toroidal propeller designs shift acoustic signatures from sharp harmonic peaks to broadband noise, defeating detection algorithms trained on the old signature.
These are not vendor weaknesses. They are the physics. Any honest acoustic drone detection system has to be designed around them, not pitched as if they do not exist.
Acoustic vs radar vs RF vs electro-optical: which counter-UAS sensor actually catches what?
The single biggest mistake European defence buyers are making in 2026 is treating these as alternatives. They are layers. Acoustic detection earns its place as the first cue. Radar refines range. Electro-optical confirms identification. Anyone selling a single-modality counter-drone system is selling you a gap.
| Sensor type | Strengths | Critical weakness | Best against |
|---|---|---|---|
| Acoustic | Passive, omnidirectional, no emissions, catches RF-silent drones | Limited range, wind sensitivity | Low-altitude Shahed, FPV, autonomous quadcopters |
| Radar | Long range, weather-resistant, precise range and altitude | Bird-drone confusion, urban multipath, reveals position | Fast-moving, larger UAS at altitude |
| RF detection | Long range, identifies protocol, locates operator | Blind to autonomous and fiber-optic drones | Commercial drones with active radio link |
| Electro-optical / infrared | Visual confirmation, classification | Narrow field of view, weather and lighting dependent | Identification once cued by another sensor |
What an acoustic drone detection network actually looks like
A working acoustic counter-UAS deployment has five components. Anyone evaluating a vendor should ask about each one.
- Sensor nodes. Ruggedized microphone arrays with piezoelectric MEMS microphones rated IP65 or higher. Capacitive microphones fail in humid outdoor conditions. The array geometry matters: 3D arrays resolve elevation, planar arrays do not.
- Edge AI processing. Onboard classification using Conformer or hybrid CNN architectures. Cloud-based processing adds seconds of latency you do not have and creates transmission signatures that reveal sensor positions. Edge inference under 100 milliseconds is the minimum bar.
- Few-shot learning capability. When Russia altered Shahed acoustic signatures, Ukraine’s Zvook system lost only 3% accuracy and recovered with minimal retraining. A system that needs weeks of retraining for every new drone variant is operationally dead.
- Multi-sensor fusion. SAPIENT protocol compatibility, currently undergoing NATO ratification, lets your acoustic sensors talk to allied radar and command systems. Proprietary protocols create the integration gaps that defeated Latvia’s response timing.
- Mobile fire team integration. Detection without a kill chain is theater. Ukraine pushes acoustic detections into the Delta situational awareness system in 12 seconds and routes them to mobile fire teams equipped with tablets and anti-aircraft guns. Without that pipeline, you have a microphone, not a defence.
The European drone wall and where acoustic detection fits
The NATO drone wall now under construction spans nearly 3,000 kilometers across Estonia, Latvia, Lithuania, Finland, Poland, and Norway. Germany’s upgraded Hensoldt ASUL system anchors the western end. Estonia’s DefSecIntel-led Eirshield platform integrates radar, cameras, RF detectors, and acoustic sensors with AI-automated decision-making. Latvia’s planned 2026 budget allocates €50 million for unmanned aerial system capabilities, a 150% increase over 2025.
Lithuania has confirmed it will deploy Ukraine’s Sky Fortress acoustic detection system in 2026. Latvia is taking a different route, building locally with Estonian and Latvian partners and seeking Ukrainian tactical expertise rather than Ukrainian hardware.
For counter-UAS vendors, this is the largest single procurement opportunity in European defence right now. For buyers, it is the moment to lock in suppliers whose sensors will integrate with whatever C2 architecture NATO settles on.
Frequently asked questions about acoustic drone detection
[faq_collapsible]
How far can an acoustic drone detector hear a Shahed drone?
Five to seven kilometers under favorable conditions, with low ambient noise and wind under 5 m/s. Range drops significantly in urban environments and high wind.
Can acoustic sensors detect autonomous drones with no radio signal?
Yes. This is the primary reason acoustic detection became critical after fiber-optic-controlled drones appeared in Ukraine. Acoustic detection is the only passive modality that catches RF-silent autonomous platforms.
Are acoustic drone detectors affected by jamming?
No. Acoustic sensors are completely passive. They emit no signal and cannot be jammed by electronic warfare. This is one of their key advantages over radar and RF detectors in contested environments.
What is the false positive rate for acoustic drone detection?
Ukraine’s Zvook system runs at 1.6% false positives after operational tuning. Western military-grade systems claim near-zero false positives in controlled conditions, though urban deployments degrade these numbers. The dominant false positive sources are birds, helicopters, motorcycles, and industrial machinery.
How much does an acoustic drone detection network cost compared to radar?
Ukraine deployed 24,000 acoustic sensors for under $5 million total. A single tactical radar covering 10 to 20 kilometers of border costs more than that. For wide-area border surveillance, acoustic is between one and two orders of magnitude cheaper per kilometer covered.
Can acoustic drone detection work in winter or rain?
Yes, but with reduced range. Precipitation adds 6 to 9 dB of ambient masking, and wet surfaces amplify nearby noise. Temperature inversions actually enhance acoustic propagation, sometimes extending detection range. Wind is a much larger problem than precipitation.
What is the difference between acoustic detection and acoustic targeting?
Detection identifies that a drone is present and provides bearing. Targeting provides weapons-grade geolocation with range and altitude. Ukrainian AReS Technologies has demonstrated phased-array acoustic targeting for FPV drones at 200 to 300 meters and Shaheds at 5 kilometers, but most acoustic systems remain detection and cueing only.
[/faq_collapsible]
What buyers should ask their counter-UAS vendor
Before signing a contract for an acoustic drone detection system, get answers to these.
- What is your detection range against a Shahed-136, a DJI Mavic, and a 14.9 dB quiet drone, each at wind speeds of 0, 5, and 10 m/s?
- How long does it take to retrain your classification model against a novel drone variant, from sample acquisition to deployed update?
- Does your sensor speak SAPIENT? If not, what is your integration path to NATO standard command and control?
- What is the latency from acoustic detection to operator alert in your reference deployment?
- What edge processor runs your inference, and what happens to detection capability if uplink to your cloud is severed?
The vendors who can answer these in specific numbers are the ones building serious counter-drone systems. The ones who answer in adjectives are not.
The bottom line for European defence buyers
Latvia did not fail because acoustic drone detection does not work. Acoustic sensors are doing exactly what physics allows them to do. The Rēzekne incident exposed a kill chain problem, not a sensor problem.
The countries getting drone wall procurement right are the ones treating acoustic detection as one critical layer in a multi-sensor fusion architecture, with edge AI that updates fast, integration paths that respect interoperability standards, and a tactical response pipeline measured in seconds rather than minutes.
If you are evaluating drone acoustic detectors for border security, critical infrastructure protection, or military forward operating bases, the right question is not whether acoustic works. It is whether your acoustic vendor understands the layered architecture, the physics constraints, and the kill chain integration that turn detection into actual defence.
About Askalon Industries
Askalon Industries designs and manufactures acoustic drone detection systems for border security, critical infrastructure, and military counter-UAS operations. Our sensors deliver passive low-altitude detection against Shahed-type loitering munitions, FPV drones, autonomous platforms, and RF-silent threats that radar and RF detectors miss.
Request a technical briefing →
Related reading
- Acoustic vs radar drone detection: when to use which
- How to build a layered counter-UAS architecture
- The European drone wall: country-by-country procurement guide
- Quiet drones and the future of acoustic detection
This article references publicly available reporting from Defense News, C4ISRNet, LSM.lv, the Latvian Ministry of Defence, and drone-warfare.com. Askalon Industries does not represent the supplier of Latvia’s eastern border acoustic detection network, which the Latvian government has not publicly disclosed.
