
(© WavebreakMediaMicro - stock.adobe.com)
Why Your Throat Snores Like a Musical Instrument, According to Science
In A Nutshell
- A new computer simulation shows that ordinary, non-apneic snoring is driven almost entirely by push-pull forces between airflow and the soft palate, not turbulence.
- The soft palate settles into a stable, repeating vibration during snoring, similar to how a clarinet reed vibrates continuously as long as air keeps flowing.
- Predicted sound levels reached about 75 decibels during inhalation and 69 decibels during exhalation, matching ranges reported in real clinical measurements.
- The findings point toward why treatments like palatal stiffening may work, and researchers plan to test how changing tissue stiffness affects the sound.
Anyone who has shared a bed with a loud snorer knows what unchecked snoring can do to sleep, sanity, and patience. An entire industry of pillows, sprays, straps, and surgeries claims to cure it, with mixed results, largely because scientists have never fully nailed down what physically produces the sound. A new simulation out of Sweden just closed part of that gap, and it points toward why some existing treatments work.
Published in the journal Physics of Fluids, the study built a 3D computer model of the upper airway to sort out how airflow, tissue, and sound actually interact during ordinary, non-apneic snoring. That distinction matters. Most snoring research targets sleep apnea, a serious and separate condition. But snoring without apnea is still exhausting for the person doing it and for whoever is stuck listening to it, and researchers still were not certain exactly how the sound gets made.
“We hope to better understand how breathing drives snoring and identify the dominant sound generation mechanisms,” said study author Peng Li, a researcher at KTH Royal Institute of Technology in Stockholm.
A Simplified Throat Model Captures Real Breathing Patterns
Touch the roof of the mouth just behind the teeth and the surface feels rigid, almost bony. That is the hard palate. Farther back, it gives way to something softer and more pliable called the soft palate, and that tissue was the focus of the KTH team’s model.
To study its physics, the researchers built a three-dimensional computer model of the soft palate and the uvula, placed inside a tube representing the throat, with nasal and oral airway channels included.
Rather than a rigid, static model, the simulation let air and tissue interact continuously. As airflow pushed on the soft palate, the tissue moved, and that movement changed the airflow around it in turn. Earlier models often assumed fixed tissue walls, making this two-way interaction a closer match for what actually happens in a sleeping throat.
Airflow in the model followed a real breathing pattern recorded from a non-apneic snoring patient, with natural rises and falls over a breathing cycle lasting about 4.1 seconds.
Credit: Li et al. (graph), Mussi Katz via flickr, licensed under CC0 (photograph)
Snoring Works Like a Musical Instrument
One of the more interesting findings involves how the soft palate vibrates. Rather than flopping passively, it settled into a stable, self-sustaining pattern that repeated consistently, similar to how a clarinet reed vibrates continuously as long as air keeps flowing.
Predicted vibration centered primarily around 2.18 Hz, with a distinct sound peak near 90 Hz, within the range reported in clinical studies of soft-palate snoring. Sound levels in the model reached about 75 decibels during inhalation and 69 decibels during exhalation, consistent with ranges reported in clinical snoring measurements.
The Sound Comes From a Tug-of-War, Not Turbulence
Scientists have long lumped snoring sound into one of a few buckets: air rushing turbulently through a narrow space, or the physical push and pull between moving air and vibrating tissue. This study settled the question for ordinary, non-apneic snoring. It is almost entirely the push and pull. Turbulence barely factored in at all, contributing less than half a percent of the total sound.
That push-pull force comes from air pressing and releasing against the soft palate every time a breath switches direction, over and over, all night long.
The Model’s Simplified Design Likely Understates Real Sound Levels
This is a simplified, idealized throat, not a scan of any real patient. It excludes throat wall flexibility, resonance from the nose or lungs, and the airway closures involved in sleep apnea. The researchers note their predicted sound levels are likely underestimates because of these simplifications, and they did not include tissue’s natural energy-absorbing properties, meaning real tissue would likely vibrate somewhat less than the model shows.
Li said the loudest sounds traced back to unsteady airflow moving across the soft palate, a finding that points toward practical fixes. “Our results suggest that reducing soft palate vibration or unsteady aerodynamic loading may help reduce palatal snoring,” Li said. “This could inform evaluation of palatal stiffening procedures or other interventions that modify tissue mechanics or airflow.”
Testing that idea directly is the team’s next move, since the current model is still too simplified to hand out specific recommendations. “Our next step is to investigate how palatal stiffness affects its vibration and the resulting snoring sound,” Li said. “By systematically varying tissue stiffness, we aim to determine how it changes oscillation amplitude, dominant frequency, airflow patterns, and acoustic source strength.”
Disclaimer: This article is based on a peer-reviewed study published in Physics of Fluids and reflects the findings and limitations described by the study’s authors. It is intended for general informational purposes and is not medical advice. Anyone concerned about snoring, sleep quality, or possible sleep apnea should consult a qualified healthcare provider.
Paper Notes
Limitations
The study uses a geometrically simplified model of the human throat rather than an anatomically realistic, patient-specific scan. The pharyngeal walls are modeled as rigid, meaning the natural compliance of real throat tissue is not captured. Structural damping of the soft palate is not included, so the vibration amplitudes predicted by the model should be treated as upper-bound estimates. The acoustic framework assumes an infinitely long, reflection-free duct, meaning real-world effects such as sound bouncing back from the ends of the airway are not modeled. Resonance effects from the nasal cavity, oral cavity, and lungs are also excluded, and the model does not address airway collapse, tissue contact, or any mechanisms specific to sleep apnea. The authors also note that the model cannot directly substitute for patient-specific clinical validation, which they describe as a goal for future work.
Funding and Disclosures
According to the paper, the project was financed by the Swedish Research Council under Grant No. VR 2020-04857, with principal investigator Professor Mihai Mihaescu. High-performance computing resources were provided by the National Academic Infrastructure for Supercomputing in Sweden (NAISS, Project No. NAISS 2024/1-13) at PDC Centre for High-Performance Computing and at the National Supercomputer Center. The project was partially financed by the Göran Gustafsson Foundation for Research, with principal investigator Dr. Marco Laudato. The authors declare no conflicts of interest.
Publication Details
Authors: Peng Li, Marco Laudato, and Mihai Mihaescu, all affiliated with the Department of Engineering Mechanics, FLOW, KTH Royal Institute of Technology, Stockholm, Sweden. | Paper Title: “Computational analysis of palatal non-apneic snoring sound generation using a simplified human upper-airway model” | Journal: Physics of Fluids, Volume 38, Article 081909 (2026) | DOI: https://doi.org/10.1063/5.0346307 | Published Online: August 18, 2026







