S. Tojo, Y. Oka, and M. Nishida (Japan)
Multimedia, Music, Parser, HPSG, Knowledge Represen tation, Grammar
In this paper, we will present an analysis system of chord progression in HPSG (Head-Driven Phrase Structure Grammar). It is said that the structure of music is somewhat similar to that of natural language sentences. When we try to write rules for chord progression in context-free gram mar, we need to add a lot of rules for trivial exceptions. However in HPSG, we can encapsulate such diversity in the feature structure of each category and can reduce the num ber of rules. In addition, we can utilize the notion of head, which is the prime constituent in lower concepts, to rep resent various hierarchical knowledge in musicology. We show that our system adequately resolves the mutual de pendency between the key and the chord progression from a given sequences of chords, and that the system outputs the candidates of plausible harmony sequences. We evalu ate the experimental results, and discuss the problems and the future extension of our method.
Important Links:
Go Back
