Can Machines Learn Indian Classical Music?


September 6, 2012


Ajay Srinivasamurthy


Universitat Pompeu Fabra (UPF)


Music Technology is a multidisciplinary field that uses the tools available to engineers to develop specific tools for understanding, analysis, and synthesis of music. Music Information Research (MIR) draws from several fields such as signal processing, machine learning, computer science, psychology, and musicology. This talk will discuss a few MIR tools in the context of Indian classical music. In specific, Multiple Viewpoint Modeling and Rhythm Modeling will be discussed in detail. Multiple viewpoint models can be used for melody continuation. Rhythm models aim to extract useful rhythm related information such as the pulse, beat, and phrase boundaries from music. Using the new set of MIR tools, we can develop several interesting applications which aim to create a novel experience with music for both musicians and the audience.


Ajay Srinivasamurthy

Ajay Srinivasamurthy is a doctoral student at the Music Technology Group, Universitat Pompeu Fabra (UPF), Barcelona, Spain. He works on rhythm modeling of Indian classical music as a part of the CompMusic project. His research interests broadly include Audio, Music and Speech Signal Processing, Psychoacoustics, Computer Music, Machine Learning, and Biomedical signal processing. Prior to joining UPF, he was a research assistant at the Georgia Tech Center for Music Technology, Atlanta. He has worked at the mobile music startup Smule Inc. in Atlanta and GE Global Research, Bangalore. He has an M.E. in Signal Processing from IISc, Bangalore and a B.Tech in Electronics and Communications from National Institute of Technology Karnataka, Surathkal.