Music Summarization from Symbolic Data




David Huron

Music Cognition Group

Ohio State University

TASK: "Play me a little bit of this piece." i.e. Find the most distinctive or representative musical excerpts automatically.
APPROACH: Using information theoretic measures to identify musical segments that have high entropy compared with other works, but low entropy within the work itself.
EVALUATION: Compare automated results with results from perceptual experiments where listeners judge the musical importance of passages.
REFERENCES: Mani & Maybury (Eds.) Advances in Automatic Text Summarization. MIT Press, 1999.