Eric Humphrey : Deep Learning in Music Informatics - New Directions for the Next Decade
As we look to the future of content-based music informatics, there is a general sense that progress is decelerating throughout the field. On closer inspection, performance trajectories across several applications reveal that this is more than just a feeling, raising some difficult questions for the discipline: why are we slowing down, and what can we do about it? This talk aims to address both of these issues.
First, common approaches to music signal analysis are reviewed in an effort to fully understand why this might be the case, and three specific deficiencies to current methods are identified: hand-crafted feature design is sub-optimal and unsustainable, the power of shallow architectures is fundamentally limited, and short-time analysis cannot encode musically meaningful structure. Acknowledging breakthroughs in other perceptual AI domains, the case is made that deep learning holds the potential to overcome each of these obstacles. Through conceptual arguments for feature learning and deeper processing architectures, it will be demonstrated how deep processing models are simply more powerful extensions of many current methods, and why now is the time for this paradigm shift.
Eric is a PhD candidate in Music Technology at the Music and Audio Research Lab (MARL) @ NYU. After earning a BSEE at Syracuse University in 2007, Eric flocked south to pursue a masters in Music Engineering Technology at the University of Miami, graduating in 2009. During the completion of his master's thesis, Eric fell in love with music informatics and New York City; as a result, he now spends his days in the Village, striving to make machines more musically intelligent. In addition to the academic pursuits of higher education, Eric is a multi-instrumentalist, has been a visiting lecturer at the University of Miami, worked as an independent contractor roles for several audio technology companies, and currently serves as the student member on the ISMIR Steering Committee.