Resumen
This paper presents a system that analyzes the time-series impression change in the acoustic signal by a unit of music phrase. The aim is to support the music creation using a computer (computer music) by bringing out composers' potentially existing knowledge and skills. Our goal is to realize the cross-genre/cross-cultural music creation. Our system realizes the automatic extraction of musical features from acoustic signals by dividing and decomposing them into â??phrasesâ? and â??three musical elementsâ? (rhythm, melody, and harmony), which are meaningful for human recognition. By calculating the correlation between the target â??target music pieceâ? and the â??typical phraseâ? in each musical genre, composers are able to grasp the time-series impression change of music media by the unit of music phrase. The system leads to a new creative and efficient environment for cross-genre/cross-cultural music creation based on the potentially existing knowledge on the music phrase and structure.