Recent work in real-time music understanding by computer

摘要: Music Understanding refers to the recognition or identification of structure and pattern in musical information. Three distinct music understanding projects initiated by author are discussed. In first, Computer Accompaniment, goal is follow a performer score. Knowledge position score as function time can be used synchronize an accompaniment live automatically adjust tempo variations. second project, it shown that statistical methods recognize location improvisor cyclic chord progression such 12-bar blues. This problem similar Accompaniment; however, much less information conveyed any one note because pitches not predetermined. The third Beat Tracking, attempts identify beats using note-onset times from performance. Parallel search techniques consider several hypotheses simultaneously, both timing higher-level knowledge integrated evaluate hypotheses.