作者: Stephen Downie , Michael Nelson
关键词:
摘要: We developed, and then evaluated, a music information retrieval (MIR) system based upon the intervals found within melodies of collection 9354 folksongs. The songs were converted to an interval-only representation monophonic fragmented t into length-n subsections called n-grams. length these n-grams degree which we precisely represent are variables analyzed in this paper. constructed “musical word” databases using text-based, SMART system. A group simulated queries, some contained errors, was run against databases. results evaluated normalized precision recall measures. Our concept words” shows great merit thus implying that useful MIR systems can be simply efficiently pre-existing text-based software. Second, study is formal comprehensive evaluation rigorous statistical analyses determine effectiveness.