Automatic Phrasing System for Expressive Perfor¬mance Based on The Generative Theory of Tonal Music

Music phrase is an ambiguous notion since it often depends on the performer's subjective view. Thus far, we have employed Director Musices (DM) for automatic expressive performance, however, segmentation of phrases has only been given manually. In order to identify phrases from an objective viewpoint, we propose to obtain them from the trees acquired by the Generative Theory of Tonal Music (GTTM). We select the usable subtrees and regard the scope of the subtrees as phrases. We introduce a test tool to generate an expressive performance, given original music data to DM together with GTTM trees, to facilitate the phrasing steps.