Parametric Analysis of Musical Vibrato in Voice and Instrument Performances

Presented by: Mingfeng Zhang

Supervised by: Professor Mark Bocko

426 Computer Studies Building

Vibrato is an important musical performance technique for voice and various musical instruments. In this paper, a signal processing framework for vibrato analysis is presented. In this framework, musical vibrato is treated as a generalized descriptor of musical timbre and several parameterized features are proposed. The author implemented signal analysis algorithms for segmenting sonic partials from musical sound and for performing amplitude and frequency tracking of these partials. Then, the author implemented a framework for musical vibrato parameterization, which extracts various vibrato parameters from the amplitude and frequency tracks. The author also has implemented extensive visualization and auralization algorithms that provide in-depth user interactions with vibrato feature tracks and their parameterizations.