Physics Student, Professor Published in Machine-Learning-Meets-Music Study

Hawley, Morrison and Morgan
(Left to Right) Hawley, Morrison and Morgan

Belmont Professor of Physics Dr. Scott H. Hawley, Joliet Junior College Professor of Physics and ASA Fellow Dr. Andrew C. Morrison, and Belmont undergraduate physics and audio engineering technology double-major Grant S. Morgan recently co-authored the research paper “Tracking Transients in Steelpan Strikes Using Surveillance Technology,” published on February 16 by the Acoustical Society of America’s journal JASA Express Letters (under “Editors Picks”).

The paper follows earlier work by Hawley and Morrison published in JASA last October in which Carribean steelpan drums were illuminated by lasers and filmed at nearly 16,000 frames per second. The researchers used neural networks to train a computer vision model that can track features on the drum surface as they evolve rapidly in time. In this new paper, the researchers improve upon prior results and even extend them to other musical instruments, such as a lyre.

Morgan’s contribution was integral to the study. As Hawley observes, “We needed to do many training runs with different parameters to find the best models for our study, and much of that work was done by Grant, who not only did the runs but consolidated and visualized the results for our selection processes.” This work was done during Morgan’s enrollment in Hawley’s course “Deep Learning and A.I. Ethics,” which is being offered again in Fall 2022 as PHY/DSC/BSA 4420.

Another key contribution of the study that Morgan helped with was “cleaning” the data used to train the neural network. Hawley added, “At Belmont, we want to raise up students who can not only perform analytics on existing data but who can think critically about how the data is obtained.”

Hawley was recently named a Fellow of the new Belmont Data Collaborative (BDC), a university-wide initiative that looks to infuse data skills into every facet of the culture and curriculum at Belmont. This includes the incorporation of different levels of data skills within each major and college as well as creating opportunities for real world projects that give students and faculty experience as well as help the social well-being of the community. Through the BDC, Belmont University will create Data Storytellers that can use data to provide meaningful insights and actionable stories.

Morgan is currently working in the world of spatial audio, assisting audio engineers that mix in Dolby Atmos. His future plans include research in the fields of audio engineering and data science to improve the immersive experience of surround sound media formats. Morgan would also like to thank Belmont University for providing him with amazing professors that helped him. He stated, “Working with Dr. Hawley allowed me to see how these interdisciplinary concepts connected in ways I never would have imagined. I’m extremely grateful that I was able to assist on this project.”

The paper is available for open access (paid for by Morrison’s NSF grant) at https://doi.org/10.1121/10.0009532. Additional materials such as movies, computer code, documentation and datasets can be found at the project’s website, https://drscotthawley.github.io/espiownage/.