Based on an observation for human brain's perception process in listening to music, a simple and useful scheme extracting the most favorable section of a digitally encoded music is proposed. The proposed scheme is experimentally shown to be advantageous over other schemes. The proposed scheme has such practical applications as an automatic system extracting the most favorable section of a song, an intelligent shuffling function for digital music playing devices, a personalized music recommendation system for online music stores, and so on(1).