Acoustic sensing is the process of detecting, measuring, and interpreting sound waves to gather information about an environment, object, or living organism. It works by converting mechanical vibrations traveling through air, water, soil, or solid structures into electrical signals that can be analyzed. The technology spans an enormous range of applications, from a doctor listening to your lungs to a fiber-optic cable detecting earthquakes along the ocean floor.
How Acoustic Sensing Works
At its core, acoustic sensing relies on a device called a transducer, which converts one form of energy into another. In this case, the transducer picks up pressure waves (sound) and turns them into electrical signals. The most familiar example is a microphone. When sound hits a microphone’s diaphragm, the vibration generates a corresponding electrical signal that can be recorded, transmitted, or analyzed by software.
Different transducer types suit different jobs. Piezoelectric transducers use crystals that produce voltage when physically deformed by a sound wave, making them ideal for ultrasound imaging and underwater sonar. Magnetostrictive transducers rely on materials that change shape in a magnetic field to generate powerful ultrasonic pulses for industrial use. MEMS (micro-electromechanical systems) sensors pack tiny vibrating structures onto silicon chips small enough to fit inside a smartphone or earbud. The consumer electronics market for MEMS acoustic sensors alone is projected to reach $1.62 billion in 2025, growing at roughly 5% annually as demand for better audio in wearable and smart devices climbs.
Passive Versus Active Sensing
Acoustic sensing systems fall into two broad categories. Passive systems simply listen. They detect sounds already present in the environment, whether that’s a whale call, a heartbeat, or the hiss of a gas leak. Active systems generate their own sound pulse and then analyze what bounces back. Sonar is the classic active system: it sends out a ping, times the echo, and calculates the distance and shape of whatever reflected it.
Each approach has trade-offs. Passive systems are stealthy and energy-efficient because they don’t emit any signal, but they depend on whatever sounds happen to be present. Active systems can probe environments on demand and extract precise measurements of distance, density, or material composition, but they require more power and can interfere with the very environment they’re measuring.
Medical Acoustic Sensing
Doctors have used acoustic sensing since the invention of the stethoscope, but modern systems go far beyond what the human ear can detect. Digital acoustic analysis can help differentiate between healthy lungs and those affected by pneumonia, asthma, COPD, pneumothorax, or pleural effusion, each of which produces characteristic sound patterns at specific frequency ranges.
Normal breath sounds fall roughly between 100 and 1,200 Hz. Abnormal sounds are more telling. Wheezes, common in asthma and COPD, have dominant frequencies between 100 and 1,000 Hz. Crackles, linked to pneumonia or fibrosis, average around 300 Hz. Rhonchi, a low rumbling sound caused by narrowed airways, sit below 300 Hz. Stridor, which signals a dangerously restricted upper airway, typically registers above 500 Hz. By analyzing these frequency signatures digitally, software can flag problems that might be missed during a quick bedside exam.
Lung function testing also uses acoustic input directly. Techniques like impulse oscillometry send low-frequency sound waves (roughly 4 to 32 Hz) into a patient’s mouth and measure how those waves travel through the airways to assess resistance and elasticity in the lungs, all without requiring the patient to perform forceful breathing maneuvers.
Wearable Health Monitors
A growing class of wearable devices uses acoustic sensors pressed against the skin to track health continuously. Small sensors placed near the neck can monitor respiratory activity and swallowing by picking up subtle vibrations on the skin surface. Accelerometer-based contact microphones worn on the chest capture heart sounds, lung sounds, and breathing rate throughout the day. Some devices assess sleep quality by analyzing snoring sounds and breathing patterns overnight.
Gastrointestinal monitoring is another frontier. Flexible, skin-attachable acoustic devices can capture bowel sounds in real time, giving clinicians a window into gut motility without invasive procedures. These wearable platforms generally combine a tiny sensor with wireless transmission, sending data to a phone or cloud system for analysis.
Distributed Acoustic Sensing With Fiber Optics
One of the most powerful forms of acoustic sensing doesn’t use microphones at all. Distributed acoustic sensing, or DAS, turns ordinary fiber-optic cables into continuous listening devices that can stretch for extraordinary distances. A DAS system fires laser pulses down a fiber and measures tiny changes in the light that scatters back. When a sound wave, vibration, or mechanical strain hits any point along the cable, it slightly deforms the glass fiber, shifting the phase of the backscattered light. By reading those phase shifts, the system reconstructs what happened and where.
The result is essentially a sensor array that spans the entire length of the cable. Commercial DAS systems can monitor fibers up to 171 km long, and because telecommunications cables already crisscross the planet, existing infrastructure can sometimes be repurposed for sensing. A 100 km fiber can detect signals up to about 1 kHz, which covers most seismic and environmental acoustic events. DAS is less sensitive to high-frequency signals than dedicated instruments like hydrophones, but its unmatched spatial coverage makes it valuable for large-scale monitoring.
Underwater and Environmental Monitoring
Acoustic sensing is the primary tool for studying underwater environments, where light travels poorly but sound carries for enormous distances. Traditional monitoring relies on hydrophones, which are essentially underwater microphones. They work well but are limited to point-based measurements, require batteries, and need manual data retrieval.
DAS is changing this. In reef monitoring studies, fiber-optic cables laid along the seafloor have detected snapping shrimp sounds in the 1.5 to 5 kHz range and various fish vocalizations between 100 and 1,000 Hz. Researchers observed that shrimp acoustic activity increased after nearby onshore lights were turned off, consistent with known low-light behavioral patterns. The same system tracked the timing and trajectory of scuba divers moving along the reef, demonstrating the spatial resolution DAS can achieve.
These capabilities matter because marine ecosystems face growing pressure from anthropogenic noise, including shipping traffic and coastal construction. Continuous acoustic monitoring helps researchers understand how noise pollution affects biodiversity and whether conservation measures are working.
Industrial Leak Detection
Pipelines carrying oil, gas, water, or chemicals are difficult to inspect physically, especially when they run underground or underwater for hundreds of kilometers. Acoustic emission sensors detect the high-frequency sound that escaping fluid produces when it forces through a crack or hole. The acoustic signature varies with the size of the leak, allowing systems to not only find a problem but estimate its severity.
Recent deep-learning models trained on acoustic emission data have achieved leak detection accuracy of 99.76% and can classify leak sizes (as small as 0.3 mm) with 99.7% accuracy. These systems work by learning the distinct acoustic patterns that different leak sizes produce, then applying that knowledge to incoming sensor data in near real time.
How Machine Learning Fits In
Raw acoustic data is often messy, especially in natural environments where wind, traffic, animal calls, and equipment noise all overlap. Machine learning algorithms sort through this complexity by extracting acoustic features from recordings and grouping similar signals together. Random forest classifiers, neural networks, and newer visualization techniques like UMAP can identify distinct sound types within massive datasets, even when background noise is heavy.
This matters for any application where you need to distinguish one sound from thousands of others: identifying a specific bird species in a rainforest recording, picking out a mechanical fault in a noisy factory, or flagging an abnormal heart rhythm captured by a wearable sensor. The combination of increasingly cheap acoustic hardware and increasingly capable software is what makes modern acoustic sensing so much more useful than a simple microphone and a pair of headphones.