Lung sounds are the noises produced by airflow through the respiratory tract, and clinicians have relied on them for over two centuries to detect conditions ranging from pneumonia to collapsed lungs. Each type of sound, whether the soft rustle of normal breathing or the musical pitch of a wheeze, originates from a specific physical event inside the airways or lung tissue. Understanding what generates these sounds, what they indicate, and where the limits of human interpretation lie reveals both the elegance and the surprising fragility of one of medicine’s oldest diagnostic tools.
Where the Sounds Come From
The sounds you hear through a stethoscope placed on someone’s chest start as vibrations created by air moving through the trachea and bronchial tree. In the trachea specifically, turbulent airflow is the dominant source: eddies in the moving air push against the tracheal wall, causing it to vibrate. Research measuring these vibrations found that tracheal breath sounds consist mainly of direct turbulent eddy pressure fluctuations, along with a smaller propagating acoustic component whose resonant frequencies correspond to the length of the upper airway.1PubMed. Measurements and theory of normal tracheal breath sounds In other words, air tumbling through your windpipe shakes the walls, and those wall vibrations are what a stethoscope picks up as sound.
As air moves deeper into progressively smaller airways, the flow becomes less turbulent and more laminar, meaning the air streams smoothly rather than swirling. That transition is why sounds heard over the peripheral lung fields are much quieter and lower-pitched than what you hear over the trachea. The lung tissue itself, along with the chest wall, acts as a filter: it absorbs higher-frequency sound waves while letting lower frequencies pass through. A systematic review of respiratory sounds in healthy people found that the highest sound frequencies are observed at the trachea, reaching up to roughly 1300 Hz during inspiration, while sounds over the lung fields are considerably lower in frequency and intensity.2PubMed. Respiratory sounds in healthy people: a systematic review
What Normal Breathing Sounds Like
Normal breath sounds fall into a few recognizable categories depending on where you listen. Over the trachea, the sound is loud and harsh, often described as hollow or tubular, with roughly equal time spent on inspiration and expiration. This is called tracheal or bronchial breathing. Move the stethoscope to the sides or back of the chest, over the lung tissue itself, and the sound softens dramatically. These vesicular breath sounds are gentle and wispy, with a longer inspiratory phase and a barely audible expiratory phase that fades quickly. Between those two extremes, over the large airways near the center of the chest, you get bronchovesicular sounds that combine features of both.
The same systematic review noted that inspiratory sounds are generally more intense than expiratory sounds when measured at most chest locations, a finding consistent across multiple studies.3PubMed. Respiratory sounds in healthy people: a systematic review The practical takeaway is that a clinician listening to your back should hear soft, low-pitched sounds that are louder when you breathe in. If the quality, loudness, or pitch of those sounds is off, or if unexpected extra sounds appear, something may be wrong.
Crackles
Crackles, sometimes still called rales, are short, sharp, popping sounds that usually show up during inspiration. They sound like the noise you hear pulling apart a strip of Velcro, or like small bubbles popping. They are among the most common abnormal lung sounds encountered in clinical practice, and they often point to fluid, inflammation, or fibrosis in the lungs.
The mechanism behind crackles is surprisingly mechanical. During normal breathing, small airways open and close as the lung inflates and deflates. When those airways snap open suddenly during inspiration, the abrupt equalization of pressure produces a brief burst of sound. Research into the physics of crackles concluded that inspiratory crackles are generated by explosive airway reopening events, while expiratory crackles arise from sudden airway closure events that are similar in mechanism but opposite in direction and much less energetic.4PubMed. Mechanism of inspiratory and expiratory crackles That energy difference is why you hear crackles far more often on the inhale than the exhale.
The timing and character of crackles help narrow the diagnosis. Fine crackles that appear early in inspiration and are brief, high-pitched, and soft tend to suggest conditions that stiffen the lung tissue, like pulmonary fibrosis. Coarse crackles that are louder, lower in pitch, and appear throughout inspiration often point to fluid in the airways, as in pneumonia or heart failure. Crackles that clear after a cough may just mean some secretions were sitting in the airways and got dislodged, which can happen in otherwise healthy people who have been lying down.
Wheezes and Their Close Relatives
Wheezes are continuous, musical sounds that usually occur during expiration, though they can appear during inspiration too. Their pitch, which can range from a low moan to a high-pitched whistle, depends on the size of the narrowed airway producing them. Asthma and chronic obstructive pulmonary disease (COPD) are the classic causes, but wheezes can also result from a foreign object lodged in an airway, a tumor pressing on a bronchus, or simple mucus plugging.
The physics behind wheezes involves flutter of the airway walls. When airways are narrowed to the point where airflow becomes restricted, the moving air causes the flexible walls to oscillate rapidly, much like the reed of a musical instrument vibrating as air passes over it. Experimental studies using collapsible tubes confirmed that these oscillations occur only when the tubes are buckled and flow-limiting, producing loud sounds whose frequency relates to the wall stiffness and the speed of airflow.5PubMed. Flutter in flow-limited collapsible tubes: a mechanism for generation of wheezes A companion theoretical model showed that a phase delay between fluid pressure and wall motion is necessary for the flutter to sustain itself, and that the critical airspeed needed to trigger it depends on the elastic properties of the airway wall.6PubMed. Flutter in collapsible tubes: a theoretical model of wheezes
Stridor is a related but distinct sound. It is a high-pitched, often harsh noise heard predominantly during inspiration, and it signals narrowing of the upper airway, in the larynx or trachea rather than deep in the lungs. Because the upper airway is a single bottleneck, stridor tends to be louder and more alarming than peripheral wheezing, and it often demands more urgent attention since it can indicate a compromised airway.
What a Pneumothorax Sounds Like (or Doesn’t)
A pneumothorax, where air leaks into the space between the lung and the chest wall, is one of the conditions where absent or diminished lung sounds carry diagnostic weight. The trapped air creates an acoustic barrier. Computational and experimental studies of sound transmission through the chest show that at low frequencies, sound still travels through a collapsed lung region almost as well as through healthy tissue. But as frequency rises, the impedance mismatch caused by the air pocket blocks sound waves progressively more, leading to a significantly weaker signal on the chest surface near the affected lung.7PubMed Central. Sound transmission in human thorax through airway insonification: an experimental and computational study with diagnostic applications Separate research found that the most pronounced reduction in transmission occurs between roughly 400 and 600 Hz.8PubMed Central. Pneumothorax effects on pulmonary acoustic transmission
This frequency-dependent behavior explains a long-standing clinical observation: when you place a stethoscope over a pneumothorax, the breath sounds are not completely gone, just muffled and notably quieter, especially compared to the healthy side. The reduction is subtle enough that in a noisy emergency room or on a moving ambulance, it can be easy to miss, which is one reason clinicians increasingly use point-of-care ultrasound alongside the stethoscope to confirm the diagnosis.
Why Lung Sounds Differ in Babies and Children
If you listen to an infant’s chest with a stethoscope, the breath sounds are louder and higher-pitched than what you would hear in an adult. This is not because babies breathe harder. It is because a smaller, thinner chest wall does much less filtering. In adults, the relatively thick layers of muscle, fat, and bone between the lung surface and the stethoscope absorb a large portion of the higher-frequency sound energy. An infant’s thin chest wall lets those frequencies pass through.
Studies comparing lung sound spectra across age groups confirmed this clearly. Median frequencies over the upper lobes measured about 280 Hz in adults, about 380 Hz in full-term infants, and about 480 Hz in premature infants, with the differences between adults and both infant groups being highly significant.9PubMed. Comparison of the lung sound frequency spectra of infants and adults A separate study found that infant spectra contain less power below 300 Hz compared with older children and adults, resulting in higher spectral frequencies overall, possibly explained by resonances of the smaller thoracic cavity.10American Journal of Respiratory and Critical Care Medicine. Lung Sound Spectra at Standardized Air Flow in Normal Infants, Children, and Adults
This matters clinically because the abnormal sounds that would be worrisome in an adult may be normal in a toddler. Bronchovesicular breathing heard over the peripheral lung fields, which in an adult might suggest consolidation from pneumonia, is often perfectly normal in small children simply because their airways are closer to the chest surface. Pediatricians learn to recalibrate their expectations, but the adjustment is experience-based, which is part of why pediatric auscultation is considered harder to teach than adult auscultation.
The Surprisingly Poor Agreement Among Clinicians
One of the less comfortable truths about lung auscultation is that clinicians often disagree about what they hear. A study assessing agreement among physical therapists found poor reliability for detecting and categorizing the quality of breath sounds, with statistical agreement scores as low as essentially zero for some categories of abnormal sounds before targeted education.11PubMed. Interrater reliability of auscultation of breath sounds among physical therapists The problem is not confined to one profession. A survey of physicians at different training levels found that while wheezes and stridor were correctly identified at high rates (above 80%), the pleural friction rub was correctly identified by only about a third of staff physicians and by single-digit percentages of residents and medical students. Normal breath sounds were also poorly recognized, with correct identification rates hovering around 11 to 21% across all groups.12BMJ Open. Influence of observer preferences and auscultatory skill on the choice of terms to describe lung sounds
Part of the problem is terminology. The vocabulary for lung sounds has shifted over the decades, with terms like rales, rhonchi, and crepitations used inconsistently across countries and even across departments within the same hospital. International efforts to standardize terminology have helped, but ambiguity persists. Another part of the problem is purely perceptual: human hearing varies, ambient noise interferes, and subtle sounds are genuinely hard to distinguish without practice. A clinician who listens to hundreds of chests a week develops a trained ear that a student simply does not have yet.
Digital Stethoscopes and AI-Assisted Listening
The traditional stethoscope has barely changed in principle since Laënnec invented the first version in 1816, a simple wooden tube he used to listen to a patient’s heart and lungs.13PubMed Central. Rene Theophile Hyacinthe Laënnec (1781-1826): the man behind the stethoscope Modern digital stethoscopes, however, represent a genuine leap. They convert chest sounds into electronic signals that can be amplified, filtered, recorded, and transmitted. This makes it possible to share a recording with a specialist thousands of miles away, replay a tricky sound multiple times, and build massive databases of annotated lung sounds for training purposes.
The real excitement is in what happens after the recording is made. Deep learning algorithms can now classify lung sounds with impressive accuracy. One study using a convolutional neural network approach reported a best classification accuracy of about 97% when using spectral representations that capture low-frequency detail well.14PubMed Central. Deep Learning-Based Classification of Common Lung Sounds via Auto-Detected Respiratory Cycles Another deep learning architecture achieved about 95% accuracy classifying six types of lung sounds from spectrograms.15Biomedical Signal Processing and Control. DeepRespNet: A deep neural network for classification of respiratory sounds These numbers compare favorably with human performance, especially for the sounds clinicians struggle with most, like friction rubs and subtle crackles.
A review of deep learning approaches for lung sound analysis noted that the field has been moving toward fully automatic systems that convert raw audio into two-dimensional spectrograms and then apply neural networks to recognize respiratory diseases or abnormal sounds without requiring a trained human ear in the loop.16PubMed Central. Deep learning-based lung sound analysis for intelligent stethoscope The potential for settings with few specialists is obvious. A portable digital stethoscope paired with a smartphone app has already been tested for COPD screening, with one system achieving about 96% diagnostic accuracy using a simple artificial neural network classifier.17Biomedical and Pharmacology Journal. AI-Integrated Digital Auscultation System for Early Detection and Monitoring of Chronic Obstructive Pulmonary Disease in Resource-Limited Settings
The Problem of Background Noise
One practical obstacle that neither analog nor digital stethoscopes fully solve is ambient noise. Lung sounds are quiet, often registering in single-digit decibels above the background at certain chest locations. In a busy intensive care unit, an ambulance, or even a noisy clinic, the sounds that matter can easily be drowned out by ventilator hiss, conversation, or equipment alarms. Research into robust auscultation systems has explored sensor arrays combined with external-facing microphones that perform adaptive noise suppression, essentially using the external microphone to identify environmental sound and then subtracting it from the chest recording in real time.18PubMed Central. Design and Comparative Performance of a Robust Lung Auscultation System for Noisy Clinical Settings When compared to several commercially available stethoscopes under simulated noisy conditions, these array-based systems maintained better signal quality.
Even without electronic help, experienced clinicians use tricks to minimize noise contamination. Asking the patient to breathe through an open mouth reduces nasal turbulence. Ensuring the stethoscope diaphragm sits flat against the skin prevents rubbing artifacts. Listening during a momentary pause in ventilator cycling gives a cleaner signal in mechanically ventilated patients. None of these fully replace a quiet room, but they push the signal-to-noise ratio in the right direction.
Veterinary Lung Auscultation and the Agreement Problem
Interestingly, the challenge of subjective interpretation is not limited to human medicine. A study of veterinarians auscultating calves for bronchopneumonia found that agreement on specific abnormal lung sounds was modest at best. Percent agreement for detecting any abnormal sound was about 78%, but when measured with more stringent statistical methods that account for chance agreement, the numbers dropped sharply, with Fleiss kappa values around 0.29 for any abnormal sound and around 0.18 for bronchial sounds.19PubMed Central. Agreement of Specific Lung Sounds Auscultation by Veterinarians for the Detection of Bronchopneumonia in Calves The finding reinforces that the core difficulty lies not in training or species-specific anatomy, but in the inherently subjective nature of listening to soft, variable biological sounds and trying to sort them into neat categories. It is a perceptual task that even well-trained ears perform inconsistently, which gives even more urgency to the development of objective, algorithm-driven tools.
When You Hear Nothing at All
Absent breath sounds are themselves a finding, and sometimes a critical one. Beyond pneumothorax, diminished or absent sounds can indicate a large pleural effusion (fluid surrounding the lung), a massive area of lung collapse (atelectasis), severe obesity where the thick chest wall absorbs too much sound, or extremely shallow breathing. In the acute setting, sudden loss of breath sounds on one side of the chest after trauma demands immediate action, because tension pneumothorax can kill within minutes if the pressure is not relieved.
The stethoscope was not designed to be a standalone diagnostic test for any of these conditions. It was designed to be a first-pass filter, a rapid bedside tool that raises or lowers suspicion and directs the clinician toward confirmatory imaging. A study comparing auscultation to chest radiography and lung ultrasonography for detecting conditions like pleural effusion, consolidation, and alveolar-interstitial syndrome found that auscultation’s diagnostic performance fell below that of ultrasonography across the board.20Anesthesiology. Comparative Diagnostic Performances of Auscultation, Chest Radiography, and Lung Ultrasonography in Acute Respiratory Distress Syndrome That does not make auscultation useless. It makes it what it always was: a screening tool that costs nothing, takes seconds, and requires no electricity. In a world where imaging is not always immediately available, especially outside well-equipped hospitals, those advantages still count for a great deal.