Tired? We Can Hear It in Your Voice #ASA190

Physical exertion affects the pitch, intensity, and temporal characteristics of speech, making speech recognition difficult for systems used by emergency response personnel and wearable devices.

PHILADELPHIA, May 14, 2026 — The “talk test” is often used as a low-tech way to measure exercise intensity: If you can easily talk or even sing, your workout is fairly light, but if conversation is difficult, you are exercising vigorously.

Physical task stress affects the coordination between breathing and speaking. Zahra Omidi from the University of Texas at Dallas studies this relationship and will present her work Thursday, May 14, at 11:15 a.m. ET as part of the 190th Meeting of the Acoustical Society of America, running May 11-15.

Group of runners participating in an outdoor race on a cloudy day, with two women in the foreground wearing athletic gear with an inset audio waveform and spectrogram showing sound intensity with a highlighted segment in red.

Vocal pitch, intensity, and pause structure are the vocal characteristics most impacted by changes in breathing and exercise. Credit: Zahra Omidi and Presidio of Monterey (CC0)

“Physical exertion directly alters respiration and phonation, and because speech shares the same respiratory system, these changes propagate into pitch, timing, and voice quality,” Omidi said.

Vocal pitch, intensity, and pause structure are the vocal characteristics most sensitive to changes in breathing and effort. Pitch and intensity both increase, while intensity also becomes less stable. Because speakers need to allocate more time to breathing, their speech rate slows down and becomes more segmented with longer and more frequent pauses.

Some of these changes might not be so noticeable to a listener, but the measurements clearly indicate a physiological difference.
“Features like pitch, intensity, and timing show clear and consistent changes, even when those differences are not immediately obvious by listening,” Omidi said. “This suggests that physical stress may operate below the threshold of perceptual salience in some cases but still induces measurable changes in the production mechanism.”

Understanding exactly how physical stress causes changes to vocal patterns can help train speech recognition systems, which often struggle with speech that differs from the average.

“Examples include emergency response, military operations, aviation under workload, and wearable voice interfaces, where people are speaking while physically active,” Omidi said. “In all these cases, speech deviates from neutral conditions due to respiratory and vocal effort constraints, leading to reduced intelligibility and system performance.”

In order to better represent real-world speech behavior, Omidi hopes researchers will adapt a more holistic view of speech variation as a reflection of a speaker’s characteristics rather than focusing solely on linguistics. Task stress is just one of the many physiological variables that can affect these variations.

“Human speech is inherently shaped by the body, and physical task stress provides a clear example of how physiological factors influence speech production,” Omidi said.

###

For more information:
AIP Media
1 301.209.3090
media@aip.org

Main Meeting Website: https://acousticalsociety.org/philadelphia/
Technical Program: https://eppro01.ativ.me/web/planner.php?id=ASASPRING2026

ASA PRESS ROOM
In the coming weeks, ASA’s Press Room will be updated with newsworthy stories and the press conference schedule at https://acoustics.org/asa-press-room/.

LAY LANGUAGE PAPERS
ASA will also share dozens of lay language papers about topics covered at the conference. Lay language papers are summaries (300-500 words) of presentations written by scientists for a general audience. They will be accompanied by photos, audio, and video. Learn more at https://acoustics.org/lay-language-papers/.

PRESS REGISTRATION
ASA will grant free registration to the in-person conference at the Philadelphia Marriott Downtown for credentialed and professional freelance journalists. If you are a reporter and would like to attend the meeting and/or press conferences, contact AIP Media Services at media@aip.org. For urgent requests, AIP staff can also help with setting up interviews and obtaining images, sound clips, or background information.

ABOUT THE ACOUSTICAL SOCIETY OF AMERICA
The Acoustical Society of America is the premier international scientific society in acoustics devoted to the science and technology of sound. Its 7,000 members worldwide represent a broad spectrum of the study of acoustics. ASA publications include The Journal of the Acoustical Society of America (the world’s leading journal on acoustics), JASA Express Letters, Proceedings of Meetings on Acoustics, Acoustics Today magazine, books, and standards on acoustics. The society also holds two major scientific meetings each year. See https://acousticalsociety.org/.

More Press Releases

AI Content Moderation Takes a Lesson from Economics #ASA190

Economic theory of attention can help understand and increase reliability of AI models searching for online hate speech.

PHILADELPHIA, May 12, 2026 — Spend enough time on the internet, and you’ll likely encounter some pretty appalling content. Hate speech tends to flourish on social media and in online communities, particularly those with little to no moderation. Even on sites with strict community standards, the volume of content makes effective moderation nearly impossible.

Large language models (LLMs) may be able to solve this problem. These AI algorithms can rapidly analyze both the content and the context of large volumes of text, filter hate speech automatically, and provide feedback to human reviewers. However, LLMs are expensive to run at scale, especially when asked to provide explanations for each piece of content they flag.

Yuan Zhao from the New Jersey Institute of Technology will present his research on creating an interpretable and low-cost method for evaluating LLMs’ hate speech classification Tuesday, May 12, at 1:30 p.m. ET as part of the 190th Meeting of the Acoustical Society of America, running May 11-15.

Man wearing glasses typing code on a Lenovo laptop in a focused programming session.

Researchers used an economic model to understand how large language models classify hate speech. Credit: Kowalski7cc on Wikimedia (CC0)

Zhao’s framework relies on the Rational Inattention (RI) model, an economic idea developed to explain human behavior. The model describes how humans act when their attention is limited and assigns a cost to that attention. According to the model, people tend to reserve their attention for high-reward decisions, spending it where it would have the greatest effect.

And while LLMs are not humans, these ideas of attention and decision-making can still be applied.

“LLMs are different from people, but we envision them as decision-makers facing some trade-off between performance and computational cost,” said Zhao. “Our approach uses the RI model as a simple yet interpretable tool to understand how LLMs make decisions.”

Zhao tested LLMs in a range of conditions to determine whether they behave like rational decision-makers. Then, he used the RI model to mimic the behavior of those LLMs, finding that it accurately predicts how LLM performance changes in different conditions.

This analysis can be utilized to guide digital communities using LLMs as part of their content moderation efforts.

“LLMs are already widely used, but there are still concerns about their reliability. Models like Rational Inattention can help make them more trustworthy by showing how their performance changes when text becomes ambiguous or intentionally disguised,” said Zhao. “This helps online platforms identify when human review is needed and where the system needs improvement.”

###

For more information:
AIP Media
1 301.209.3090
media@aip.org

Main Meeting Website: https://acousticalsociety.org/philadelphia/
Technical Program: https://eppro01.ativ.me/web/planner.php?id=ASASPRING2026

ASA PRESS ROOM
In the coming weeks, ASA’s Press Room will be updated with newsworthy stories and the press conference schedule at https://acoustics.org/asa-press-room/.

LAY LANGUAGE PAPERS
ASA will also share dozens of lay language papers about topics covered at the conference. Lay language papers are summaries (300-500 words) of presentations written by scientists for a general audience. They will be accompanied by photos, audio, and video. Learn more at https://acoustics.org/lay-language-papers/.

PRESS REGISTRATION
ASA will grant free registration to the in-person conference at the Philadelphia Marriott Downtown for credentialed and professional freelance journalists. If you are a reporter and would like to attend the meeting and/or press conferences, contact AIP Media Services at media@aip.org. For urgent requests, AIP staff can also help with setting up interviews and obtaining images, sound clips, or background information.

ABOUT THE ACOUSTICAL SOCIETY OF AMERICA
The Acoustical Society of America is the premier international scientific society in acoustics devoted to the science and technology of sound. Its 7,000 members worldwide represent a broad spectrum of the study of acoustics. ASA publications include The Journal of the Acoustical Society of America (the world’s leading journal on acoustics), JASA Express Letters, Proceedings of Meetings on Acoustics, Acoustics Today magazine, books, and standards on acoustics. The society also holds two major scientific meetings each year. See https://acousticalsociety.org/.

More Press Releases

“Listening” to Arctic sea ice: Using fiber optic cables to track when it might break

Junsu Jang – junsu.jang@whoi.edu

Applied Ocean Physics & Engineering, Woods Hole Oceanographic Institute, Woods Hole, MA, 02543, United States

Maddie Smith
Gil Averbuch

Popular version of 1aSP – Sea ice property inversion using distributed acoustic sensing on Arctic landfast ice
Presented at the 190th ASA Meeting
Read the abstract at https://eppro01.ativ.me/web/planner.php?id=ASASPRING2026

–The research described in this Acoustics Lay Language Paper may not have yet been peer reviewed–

Diagram showing an ice radar on a shore transmitting signals to buoys and receivers across a 5 km stretch of icy water with fiber optic cables connecting them.Fig. 1: Schematic of landfast sea ice and the field setup used in this study. Landfast sea ice is attached to the seafloor near the coast, often anchored by grounded ridges (shown here). A fiber optic cable (blue line) is laid along the snow-ice interface and acts as a series of sensors that “listen” to vibrations in the ice. The figure is not to scale. (Figure by Maia LeDoux and the Applied Physics Laboratory Graphics Department; cropped and annotated by the authors to show the fiber optic cable.)

Arctic landfast sea ice is the ice attached to the seafloor near the coast (see Fig. 1). It plays an important role in ocean–atmosphere interactions and supports local communities, wildlife, and coastal stability. As the climate warms, knowing when this ice might crack or break away is increasingly important for both community safety and coastal protection.

Studying landfast ice is difficult. Researchers often have to drill through thick ice or drag along heavy instruments across large areas. Satellites help, but clouds and limited coverage can leave gaps. We need a way to continuously “listen” to the ice over long distances.

Our solution uses a technology called distributed acoustic sensing. It turns a standard fiber optic cable, similar to what brings internet to homes, into hundreds of vibration sensors. Instead of placing many separate instruments, one cable can measure motion along its entire length with high detail.

In 2025, we installed a 2-kilometer-long cable across landfast sea ice in Arctic Alaska (see Fig. 2). A custom sled cut a shallow trench in the snow and ice, laid the cable, and covered it. This setup effectively created about 600 sensors recording vibrations 500 times per second.

Satellite view of a snowy Arctic coastline with a red line marking a path from inland towards the water's edge.Fig. 2: Satellite image of the landfast sea ice showing the 2-kilometer-long fiber optic cable (red line). The cable extends from near the coast out across the ice. Image taken on May 26, 2026. (Image © Planet Labs PBC, CC BY-NC-SA 2.0; labels, cable layout, and axes added by the authors.)

What did we hear? We detected waves traveling through the ice, generated by ocean swells offshore (see Fig. 3). The ice behaves like a thin floating plate sitting on the water, bending as waves pass underneath. By analyzing these motions, we can estimate how stiff or “bendy” the ice is and how much stress it is under from waves and wind.

Color-coded strain variations along a 1.2 to 2.0 km cable over 60 seconds on 2025-05-05 UTC, showing alternating red and blue wave patterns.Fig. 3: Example of measurements from the fiber optic cable. The horizontal axis shows time, and the vertical axis shows distance along the cable (farther from shore upward (see Fig. 2). Red and blue bands indicate the ice stretching and compressing as ocean waves pass underneath, causing the ice to bend. By analyzing these patterns, we can estimate how stiff the ice is and how it responds to waves.

This information will help answer key questions: How thin or weak does the ice need to be before it breaks? What role do waves and wind play? Ultimately, this can improve predictions of “breakout” events, when large pieces of ice detach, and seasonal breakup.

This work is a collaboration with a broader effort, the Arctic PISCES project, to better observe, understand and predict the ocean-ice-atmosphere system in Arctic coastal and inner-shelf regions. With continued monitoring, fiber optic sensing could become a powerful new way to track the stability of Arctic sea ice.

Listening to Liberty: How Modal Analysis Sheds Light on the Sound of the Liberty Bell

Sean Collier – smc604@psu.edu

The Pennsylvania State University Applied Research Lab, University Park, PENNSYLVANIA, 16804, United States

Jonathan Young
Aaron Stearns

Popular version of 4aMU5 – Modal Analysis of a Liberty Bell Replica
Presented at the 190th ASA Meeting
Read the abstract at https://eppro01.ativ.me/web/planner.php?id=ASASPRING2026

–The research described in this Acoustics Lay Language Paper may not have yet been peer reviewed–

More often than not, what crosses your mind when hearing “Liberty Bell” is not the chime or musicality of the bell; rather, it is likely the infamous crack through its left-hand side. On this 250th anniversary, it is worthwhile to pause in its silence, and consider how it might have sounded when newly cast in 1751.

Predicting the sound of a bell feels, at first pass, relatively straightforward. The bell has a simple geometry, which is important for understanding how its vibration patterns radiate sound to your ears. The bell is made of bronze, which is essential for predicting the exact pitches at which the bell will sound, as well as how long it rings. These parameters, considered together, define this action of vibrations along the surface pushing air into oscillation, leading to the radiation of sound and the bell’s characteristic chime. The physics are well understood, but the Liberty Bell, in particular, is not. That is, surprisingly little exists in the way of geometry and material composition for the bell, which makes it a particularly challenging example.

Large bell mounted on a sturdy black metal frame inside a building.Figure 1. Photo of the Liberty Bell replica at Penn State Behrend

Simply using the meager information available online produces, frankly, a terrible sounding bell that in presence does not match the silhouette of the iconic bell. Rather than toy with parameters until things seemed “right,” we sought out a replica to measure its geometry and vibroacoustic response, so to calibrate our prediction. Modal Analysis, the act of exciting a structure to understand its vibration patterns – called modes – as well as its frequencies at which the modes vibrate, is a common tool used to isolate this information in a meaning and practical way. Some results from this modal analysis are compared in Figure 2, showing the measured vibration patterns for the replica to classical results from Rossing and Perrin [1].

3D visualizations of a bell showing vibration modes labeled hum, fundamental, and tierce with corresponding 2D outline diagrams below each.Figure 2. Comparison of the first few vibration patterns between the replica and theory

Knowing the modes and frequencies was only half the effort, though, as we noted that the geometry defines so much of the sound that we eventually perceive. Indeed, small changes to the geometry could alter the prediction considerably. To have the model be as close to truth as possible, a 3D scan of the replica was done to produce a geometry – making it likely the most accurately modeled cast bell to ever exist! Once the geometry, frequencies, and modes are in place, the prediction could be tuned so to back out the bell’s material properties – “Bell Bronze”, intrinsic to the distinct ring of bells.

Large bronze bell with wooden yoke and metal supports, shown from multiple angles including detailed 3D renderings.Figure 3. 3-dimensional scan of the replica to define the cross-section and model geometry

Through modal analysis of this replica, we were able to tune a predictive model of the bell to match the measured vibroacoustic response. Beyond the pretty shapes, the analysis tells us how pitches in the chime relate in strength and in time, illuminating the evolution of the sound over time and adding scientific context to something so often overlooked in the story of the Liberty Bell.

References:

[1] Rossing, Thomas D., and Robert Perrin. “Vibrations of bells.” Applied Acoustics 20.1 (1987): 41-70

Neural Network Helps Detect Gunshots From Illegal Rainforest Poaching

Proposed model filters and verifies signals among a network of microphones to reduce false positives for gunshot detectors #ASA_ASJ2025 #ASA189

HONOLULU, Dec. 2, 2025 — Wildlife poaching remains a major conservation concern. Technological advancements have enabled webs of acoustic sensors to be deployed throughout rainforests, creating the possibility of real-time alerts to the sounds of gun-based poaching.

But the belly of the rainforest is loud, and sorting through a constant influx of sound data is computationally demanding. Detectors can distinguish a loud bang from the whistles, chirps, and rasps of birds and bugs. However, they often conflate the sounds of branches cracking, trees falling, or water dripping with gunshot noises, resulting in a high percentage of false positives for gunshot detectors.

Naveen Dhar, along with collaborators from Cornell University’s K. Lisa Yang Center for Conservation Bioacoustics and Elephant Listening Project, aimed to develop a lightweight gunshot detection neural network that can accompany sensors and process signals in real-time to minimize false positives.

Dhar will present his model Tuesday, Dec. 2, at 3 p.m. HST as part of the Sixth Joint Meeting of the Acoustical Society of America and Acoustical Society of Japan, running Dec. 1-5 in Honolulu, Hawaii.

A photo of a male forest elephant captured near the site where some of the gunshot recordings were taken. Credit: Anahita Verahrami, Elephant Listening Project

A photo of a male forest elephant captured near the site where some of the gunshot recordings were taken. Credit: Anahita Verahrami, Elephant Listening Project

The model works with autonomous recording units (ARUs), which are power-efficient microphones that capture continuous, long-term soundscapes. The proposed system utilizes a web of ARUs deployed across the forest, each performing real-time detection, with a central hub that handles more complex processing.

An initial scan filters all audio for “gunshot likely” signals and sends them to the ARU’s microprocessor, where the lightweight gunshot detection model lives. If confirmed as a gunshot by the microprocessor, the ARU passes the information to the central hub, initiating data collection from other devices in the web.

By determining if other sensors also hear a “gunshot likely” noise, the central hub then decides whether the event was a true gunshot or a potential false positive. If it determines a true positive, the central hub collates audio files from each sensor, allowing it to pinpoint the location of the gunshot and alert rangers with coordinates for immediate poaching intervention.

“Down the road, the device can be used as a tool for rangers and conservation managers, providing accurate and verifiable alerts for on-the-ground intervention along with low-latency data on the spatiotemporal trends of poachers,” said Dhar.

He plans to expand the model to detect the type of gun that fires each gunshot and other anthropogenic activities, such as chainsaws or trucks, before field-testing the system, which is currently under development.

“I hope the device can coalesce with Internet of Things infrastructure innovations and cost reduction of materials to produce a low-cost, open-source framework for real-time detection usable in any part of the globe,” said Dhar.

Contact:
AIP Media
+1 301-209-3090
media@aip.org

——————— MORE MEETING INFORMATION ——————–

Main Meeting Website: https://acousticalsociety.org/honolulu-2025/
Technical Program: https://eppro02.ativ.me/web/planner.php?id=ASAASJ25

ASA PRESS ROOM
In the coming weeks, ASA’s Press Room will be updated with newsworthy stories and the press conference schedule at https://acoustics.org/asa-press-room/.

LAY LANGUAGE PAPERS
ASA will also share dozens of lay language papers about topics covered at the conference. Lay language papers are summaries (300-500 words) of presentations written by scientists for a general audience. They will be accompanied by photos, audio, and video. Learn more at https://acoustics.org/lay-language-papers/.

PRESS REGISTRATION
ASA will grant free registration to credentialed and professional freelance journalists. If you are a reporter and would like to attend the meeting and/or press conferences, contact AIP Media Services at media@aip.org. For urgent requests, AIP staff can also help with setting up interviews and obtaining images, sound clips, or background information.

ABOUT THE ACOUSTICAL SOCIETY OF AMERICA
The Acoustical Society of America is the premier international scientific society in acoustics devoted to the science and technology of sound. Its 7,000 members worldwide represent a broad spectrum of the study of acoustics. ASA publications include The Journal of the Acoustical Society of America (the world’s leading journal on acoustics), JASA Express Letters, Proceedings of Meetings on Acoustics, Acoustics Today magazine, books, and standards on acoustics. The society also holds two major scientific meetings each year. See https://acousticalsociety.org/.

ABOUT THE ACOUSTICAL SOCIETY OF JAPAN
ASJ publishes a monthly journal in Japanese, the Journal of the Acoustical Society of Japan as well as a bimonthly journal in English, Acoustical Science and Technology, which is available online at no cost https://www.jstage.jst.go.jp/browse/ast. These journals include technical papers and review papers. Special issues are occasionally organized and published. The Society also publishes textbooks and reference books to promote acoustics associated with various topics. See https://acoustics.jp/en/.

More Press Releases

Listening to the Placenta to Detect Pregnancy Complications Early

Farah Deeba – fdeeba@charlotte.edu

University of North Carolina at Charlotte
9201 University City Blvd Charlotte
Charlotte, NC, 28223
United States

Additional Authors: William Hempstead, Hamid Moradi, Mekdes Bezabh, Robert Rohling

Popular version of 2aBAb3 – Quantitative Ultrasound Characterization of the Human Placenta for Detection of Placenta-Mediated Pregnancy Complications
Presented at the 189th ASA Meeting
Read the abstract at https://doi.org/10.1121/10.0040285

–The research described in this Acoustics Lay Language Paper may not have yet been peer reviewed–

Many pregnancy complications begin with subtle problems in the placenta, the organ that supplies oxygen and nutrients to the baby. Can sound waves reveal these hidden problems before they lead to serious health risks? Our research shows that they can: a simple ultrasound scan may help identify placental problems much earlier than today’s clinical methods.

When the placenta fails to provide adequate support, a condition known as placental insufficiency, pregnancies are at higher risk for preeclampsia in the mother, and growth restriction and oxygen deprivation (hypoxia) in the baby. Because these conditions often show no early symptoms, clinicians need a safe, accessible way to assess placental health during routine prenatal care.

To address this need, researchers at UNC Charlotte and the University of British Columbia are studying a technique called quantitative ultrasound (QUS). QUS analyzes the raw sound echoes returning from the placenta during an ultrasound scan. These echoes contain detailed information about tissue structure. When the placenta begins to show signs of insufficiency, its acoustic “signature” changes. QUS can detect these subtle changes that are not visible on a regular ultrasound image. Video 1 shows the raw signals (right) corresponding to an in utero placental image (left) acquired during a third-trimester ultrasound scan. These raw signals contain the acoustic information that QUS analyzes to detect early changes in placental health that may not be visible on a standard ultrasound image.

Video 1. Ultrasound data collected from a 3rd trimester placenta, along with the raw sound-wave signal (RF signal) that QUS analyzes.

To test whether these acoustic signatures reflect real structural differences, we first applied QUS on placentas collected after delivery. Using QUS, which measures how placental tissue absorbs and weakens sound, scatters the echoes, and what are the average sizes of the tissue structures, we found clear differences between healthy and diseased placentas. When these measurements were entered into a simple prediction model, the tool correctly distinguished healthy and diseased placentas with high accuracy showing that QUS can capture structural changes linked to placenta-mediated diseases: preeclampsia and small-for-gestational-age.

Encouraged by these findings, we evaluated QUS during pregnancy, conducted within the Wellcome Leap In Utero Consortium, an international effort to understand and prevent stillbirth. In this in utero study, we scanned pregnant participants in the USA, Canada, UK and Uganda, and analyzed the acoustic patterns of their placentas. QUS measurements were able to identify pregnancies in which babies later experienced oxygen-related distress and were especially accurate when these complications were linked to placental abnormalities confirmed after birth. Figure 2 shows how raw (RF) signals are transformed into a color-coded QUS map, making subtle differences in placental tissue easier to see and compare. These findings suggest that QUS could help clinicians recognize early signs of risk and monitor pregnancies more closely.

Figure 1: Raw sound waves collected during the scan are transformed into a color-coded quantitative ultrasound (QUS) map that highlights acoustic differences within the placenta during pregnancy

Figure 1: Raw sound waves collected during the scan are transformed into a color-coded quantitative ultrasound (QUS) map that highlights acoustic differences within the placenta

Because QUS uses the same sound waves and equipment already found in clinics, it can be integrated into handheld or portable ultrasound devices, making it practical for hospitals, local clinics, and resource-poor settings, where advanced imaging is not available. This flexibility gives QUS the potential to support more equitable prenatal care worldwide. As we continue refining the technology, our goal is to develop a fast, affordable tool that detects placental insufficiency early enough to improve pregnancy outcomes everywhere, including communities with limited access to specialized medical care.