If We Had a HAMR…
Electrical Engineering Professor Dan Ellis’ Laboratory for the Recognition and Organization of Speech and Audio (LabROSA) held its first music and audio hackathon June 28 to 30, 2013. Hacking Audio and Music Research (HAMR) drew more than 30 participants with diverse backgrounds and experience levels from a broad range of universities, organizations, and even a high school. They spent the weekend in the School’s Interschool Lab working on new research projects centered on the development of new techniques for analyzing, processing, and synthesizing audio and music signals.
“I was delighted to have so many researchers from so many different places coming together to work so hard,” says Ellis. “There was a lot of substantial research accomplished, and I'm looking forward to seeing how this work impacts the wider research community.”
The hackathon was organized by Colin Raffel, a PhD student whose advisor is Ellis, with support from the School’s electrical engineering department. "As a prototype for research hackathons, HAMR was a huge success,” Raffel observes. "After attending a few music hackathons, I realized the 'rapid prototyping' model could also be applied to research problems. The idea was to allow people to do the crucial initial work that determines whether the work is viable or not."
Ellis and Raffel note that HAMR's focus was different from that of a typical hackathon: the winning projects were chosen based on their research potential, not for their commercial viability.
A series of keynote talks from Changxi Zheng, assistant professor of computer science at Columbia Engineering; Douglas Repetto, assistant professor of professional practice in visual arts at Columbia; and Juan Pablo Bello, associate professor of music technology at New York University, kicked off the hackathon Friday night. On Saturday, the hackathon began in earnest, and continued on Sunday. At the end of the weekend, participants gave brief presentations on their projects and wrote a short précis on HAMR's "proceedings" wiki.
HAMR hammers were awarded by popular vote to Andy Sarroff (Dartmouth College) for best code for deepAutoController; Finn Upham (NYU) for best documentation for Respiration Phase Coordination; and Tom Stoll (Dartmouth), Irina Likhtina (Columbia Engineering, BS’05), and Dennis Heihoff for best hack/research direction for Synaesthesia.
Raffel says that most of the participants plan on continuing work on their projects in hopes of eventually submitting them for publication, and that planning is already underway for the next iteration of HAMR, slated for 2014.
"We really didn't know what to expect, or even who would show up," admits Ellis, "but HAMR was everything we hoped for. We've found that intensely focused work over a weekend is enough to get an idea from something just sitting at the back of your mind to a real and viable project with momentum. It's a great lesson, and I'm really looking forward to doing it again."
-by Holly Evarts