Prosodic Modeling of Nagauta Singing and Its Evaluation

Nagauta (長唄) is one of the classical styles of Japanese singing characterized by original and unique prosodic patterns. Abrupt and sharp changes of F0 are often observed and they induce simultaneous abrupt changes of power. It is very interesting that the F0 increases are synchronized with the power decreases. In our previous study, we proposed two models to synthesize the unique prosodic patterns from standard scores. Both of the F0 and power patterns are acoustically modeled as damping oscillations realized by second-order systems. In this work, the proposed models are evaluated from two viewpoints. The first evaluation is rather elementary, where three very simple musical scores are used to test the F0 and power models. In the second evaluation, a standard score of a real song is adopted to synthesize its Nagauta prosodic patterns. Here, a well-known Japanese song of “Furusato” (my old hometown) is used and the results indicate high validity and effectiveness of the proposed models.