Saving Birds One Song at a Time
Using artificial intelligence and an extensive network of acoustic recorders, Pitt doctoral student Tessa Rhinehart is advancing wildlife conservation by studying bird vocalizations.
For Rhinehart, the path to bioacoustics research began when she was an undergraduate student. "I was a naturalist for the summer at a state park near my house, and I volunteered to help with a bird research project on the weekends," Rhinehart explains. "I got to hold a bird and see all this diversity of birds that was around me. I didn't realize that there are dozens of species just in my backyard."
This hands-on experience sparked a fascination that would shape Rhinehart’s academic trajectory. Rhinehart had begun her undergraduate career pursuing a degree in mathematics but added a biology major immediately after that summer.
“Math was what I initially planned to do because I always liked it,” recalls Rhinehart. “It turned out that math has really diverse applications beyond just doing calculations. Studying math helped me discover a love for computer science and for statistics and mathematical modeling. Having all these skills together, I can ask research questions without computation being a barrier.”
This interdisciplinary background perfectly positioned Rhinehart for the emerging field of bioacoustics, where biology meets big data. After graduation, Rhinehart's search for opportunities that combined her computational skills with her passion for wildlife led her to apply for a technician role in Associate Professor Justin Kitzes's lab in the Dietrich School’s Department of Biological Sciences.
The job description—using computer science and sound to study bird populations—"ticked all of my boxes," Rhinehart says. "I couldn't imagine a more perfect fit."
Rhinehart loved the work in the Kitzes lab so much that she decided to pursue a PhD. Though she explored options at several institutions, she ultimately chose to remain at Pitt because of two crucial resources: the Kitzes lab's extensive acoustic equipment (over 2,000 recorders) and the University's Center for Research Computing (CRC).
"It's just not feasible to do my research on a laptop," Rhinehart admits. "I really need a computing cluster to do it, and the fact that the CRC provides those resources to us is what makes the research possible."
Rhinehart's research exemplifies how technology is revolutionizing wildlife monitoring. Traditional field methods for studying birds typically involve brief, periodic visits to observation sites, but this approach is time-consuming and can miss species that aren't active or present during those short visits. Bioacoustic monitoring allows researchers to collect more data than ever before.
"In the field, you don't get as much data, and you can maybe only visit a site twice or three times a year at the scale that we're doing this work. In contrast, our equipment is recording continuously during the best time of day for these birds to sing—and is doing it for six weeks straight," says Rhinehart.
This type of continuous monitoring is possible thanks to affordable acoustic sensors and powerful AI models.
Says Rhinehart, "Bioacoustics has been around for a long time, and originally, it was used to study animal vocalizations and communication. People had created simple sound processing detectors that worked decently for this, but the advent of an AI model called a convolutional neural network—the same thing Google uses to tell you if there's a cat in your photos—works super well for identifying sounds on spectrograms, which are image representations of sound.”
These technological advancements allow researchers to process years’ worth of continuous audio that would otherwise be impossible to listen to and analyze. This allows Rhinehart to identify specific bird species across large datasets and ask questions about bird populations over time. Rhinehart's work has significant implications for conservation efforts, for example, acoustic monitoring can help assess whether restoration efforts are achieving their goals. This approach facilitates adaptive management, where conservation strategies can be refined based on quantifiable feedback.
"One of the largest barriers to preventing biodiversity loss is that wildlife is losing habitat," Rhinehart says. "Preserving this habitat and restoring degraded habitat to bring back species that can't thrive in current conditions is a key goal for biodiversity conservation."
Field research can produce surprising and encouraging results. During a recent field season testing a "moth photo booth," Rhinehart heard the call of an Eastern Whip-poor-will, a nocturnal species whose population has declined by 60% over the past 50 years.
“I was excited to hear a Whip-poor-will, especially because the largest part of their diet is moths, which is what I was studying,” says Rhinehart.
When Rhinehart later analyzed recordings from multiple sites, she found that there were Whip-poor-wills at almost every site she had deployed a recorder.
Observes Rhinhart, “That was really exciting to see that the population still has a stronghold in that region.”
As her doctoral studies progress, Rhinehart is excited about the methodological advances in her field and the conservation work she is contributing to. Most recently, Rhinehart and her colleagues published work that demonstrates the potential to identify individual animals through their calls, similar to how human voices are distinguishable from one another.
"Using acoustic recorders deployed over long periods of time, we can see whether or not certain areas have high survival rates of individuals," Rhinehart says. "We can count the number of individuals that use one environment versus another just by measuring how many different individuals are detected on the recorder at any given place and time."
While Rhinehart acknowledges that there is a lot of uncertainty in what she will pursue after graduation, her unique combination of computational skills, biological knowledge, and passion for wildlife represents a new generation of conservation scientists who are harnessing technology to protect biodiversity. Ultimately, she hopes to continue her work in conservation and research.
“A lot of my PhD has been focused on building all these methods and I do love using the methods,” says Rhinehart. “But my number one love is for the birds. I'm so excited to apply these methods to further bird conservation.”
Return to the July 2025 issue »