David Campos Anchieta recently received the award for a paper he presented at the ASA meeting in Sydney, Australia
/filters:quality(90)/prod01/production-cdn-pxl/media/umassdartmouth/news/2024/IMG_4334-1920X1080-1-1200X675.jpg?text=945+WebP)
Electrical engineering PhD student David Campos Anchieta recently won the Acoustical Society of America (ASA) Best Student Paper Award for a paper he presented at the ASA's Sydney meeting in December 2023. Anchieta received a cash prize of $300 and a certificate celebrating his win. Anchieta's research was part of a project sponsored by the Office of Naval Research called Universal Adaptive Beamformers.
"My paper's title was 'Robust power spectral density estimation via a performance-weighted blend of order statistics,'" Anchieta said. "In short, the power spectral density (PSD) tells how loud each frequency is in a piece of sound. We want to estimate PSD because it carries information about the environment where the sound was recorded.
"For example, if you leave a microphone recording of sounds right below the surface of water, the PSD would be different when it's raining, dry, or windy. However, the environment may have other loud sounds from diverse sources, such as animals, boats, etc. Those sounds are outliers in the acoustic data that make it harder to estimate the PSD of the background noise. It's like trying to figure out the average income of a group of people in a classroom, but Jeff Bezos sneaked in, and now your average is much higher than it should be.
"By using order statistics (such as median, for example) we can avoid the bias caused by those outliers. But there are several ranks of order statistics to choose from between the minimum and the maximum. To choose the best rank for each moment, we borrowed an idea from a technique used in signal processing called universal adaptive beamforming. The idea is to evaluate the performance of each of the rank order statistics and choose the ones that perform best."
"David created a robust new technique for estimating the frequency content of the undersea soundscape, known as the power spectrum," explained Chancellor Professor John Buck. "Previous approaches to estimating power spectra suffered from bias created by noisy outliers such passing ships and marine animals, which David's technique successfully discards. His results are similar to the animated graphic equalizer we all know from music apps on our phones, but for underwater sound recordings instead of music. Accurately estimating the power spectrum in undersea recordings is critical for detecting rainfall from undersea recordings, tracking spawning fish populations, as well as many defense applications."
Buck and Anchieta both presented at the ASA meeting in Sydney in December 2023. "It was a great experience traveling to Sydney for this meeting," Anchieta said. "The city is really beautiful, and the meeting was great. I was able to meet colleagues from other institutions and get some feedback on my own project."