This September, Casey began a Google-sponsored project called "Search by Groove" that identifies recurring rhythmic patterns in music recordings by examining the multiple layers of sound in a given song.
"Music is a part of human life, so machine listening is very useful," Qingyuan Kong, a third-year PhD student working on the "Search by Groove" project, said.
This technology is more accurate and better tailored to the listener's needs than other popular music services, according to Casey, who is the music department chair and head of Bregman Music and Audio Research Studio at Dartmouth. Pandora, a music-listening engine that suggests songs based upon similar genres, is often criticized for repeating songs on a given Pandora channel when it is played for a prolonged period of time. "Search by Groove," which would draw on a bigger song database and thus offer greater variety, will use the technology developed by Casey to more accurately match songs.
Collaborating with a large company like Google has been critical to cataloging such a wide variety of songs, according to Andrew Sarroff, a first-year PhD student and Neukom Graduate Fellow who is one of the students working with Casey on the project.
"This is identifying a challenge that many similar research labs face," Sarroff said. "There are millions and millions of songs out in the clouds, and the kind of work we do requires scale."
Casey is also working with fMRI technology to monitor brain response to music in hopes of finding brain reaction patterns to similar music. The technique examines specific features of music, such as rhythm, harmony and vocals, according to Casey.
"[fMRIs] are important because though we're sending machines out to do the listening to figure out what media might be related, we need to understand how to tell the machines what to listen for," Casey said.
Jessica Thompson, a first-year masters student studying fMRI, said she has always been fascinated by her own response to music and how music impacted others around her. Thompson set out to research the mechanism by analyzing brain images to see which parts of the brain are affected by music.
"Music directly influences areas of the brain that control rhythmic behavior and emotion," Thompson said.
Those patterns are then mapped out and translated to machine-readable forms so that machine systems can interpret music in ways that are relevant to humans, she explained.
This October, Casey receieved a grant from the National Endowment for the Humanities for a project called the Audio-visual Cinematic Toolbox for Interaction, Organization and Navigation, which focuses on the computational analysis of film and audio materials. Casey is working with film and media studies professor Mark Williams to analyze over 100 movies, looking for patterns within and across genres and directors.
With the ACTION project, Casey said he hopes to shed new light on film criticism and theory by highlighting previously hidden film archetypes. The Dartmouth team has already explored Hitchcock films and found interesting properties in Hitchcock's use of sequential color changes that could potentially give insight into his directing technique, according to Casey.
The research team hopes to develop software that will facilitate others doing research in the same area, Casey said.
Casey is also collaborating with mathematics professors Dan Rockmore and Scott Pauls on the Music Archive Content Curation by Algorithms project, which consists of a partnership with British Broadcasting Corporation. Casey, Rockmore and Pauls are looking to conduct an analysis of a subset of a BBC archive that contains 10,000 to 12,000 recordings of pop songs made between 1967 and 2004 and convert the data into machine-readable form.
The collaboration will guarantee Casey access to an immense audio archive for research, while the BBC will benefit from Casey's cutting edge music-listening technology.
Casey, who studied music in college, quickly realized he needed a fairly deep understanding of technology to advance his musical development and therefore began to explore music's more technical aspects, he said.
"Dartmouth is uniquely positioned to support interdisciplinary collaboration between various departments such as music and science," Casey said. "Its small size is an advantage in many ways."
The Bregman program studio at Dartmouth dates back to the 1960s and has a long history of being at the cutting edge of music and technology, according to Casey.
"We all come from really different backgrounds and have something completely different to bring to the table," Sarroff said. "It's invigorating work and really fun. Oh, and we have an intramural soccer team. We intend to win."
