On Tuesday June 21st at 12:00h in room 55.309, Parag Chordia, from Georgia Tech, give a seminar on "Computational Modeling of North Indian Classical Music".
Abstract: This talk will present an overview of computational and cognitive research on North Indian classical music (NICM). NICM is one of the world's most sophisticated musical traditions, with an emphasis on melody and rhythm. The talk will discuss experiments in raga recognition, the melodic abstraction around which almost all NICM is organized, with an emphasis on features and classification methods that have proved effective. NICM is characterized by flowing melodies, with a variety of ways of connecting and ornamenting notes. I will describe recent work on expressive pitch modeling based on parametric modeling. Tabla, a pair of drums played with the hands, is synonymous with NICM, and is one of the most developed solo percussion traditions in the world, possessing a sophisticated repertoire of compositions and improvisation techniques. Tabla is based on timbral patterns that are often arranged hierarchically leading to complex patterns. I will describe sequence models that we have used for tabla recognition and prediction from both audio and symbolic scores. Additionally I will present a generative tabla system, tabla gyan, that attempts to model the creative process of tabla improvisation. The theory and practice of NICM is based on the idea that raags predictably evoke certain emotions. In several large-scale surveys, we show that such emotions are recognized in nearly the same way by enculturated and novice listeners, and that the valence of responses is closely related to the scale type. In a follow-up experiment, we show that these responses may be due to comparison with established scale norms, with lowered scale degrees leading to more negatively valenced emotions, and raised scale degrees leading to more positively valences emotions.