Abstract
AbstractThe advancement in the field of music signal processing insists on effective music information retrieval (MIR) techniques. Query by humming (QBH) system is one of the active research areas under MIR. In this paper, we propose a QBH system for automatically retrieving the desired song based on humming query and human perceptual features. In the proposed system, five perceptual features corresponding to four perceptual properties are extracted. Further, the temporal relationship of these features is estimated through the transfer entropy (TE). The trajectory of TE of the target music database is analyzed to find the match for humming query. Series of experiments are conducted to evaluate the effectiveness of the system with 1,200 songs target database and 200 humming queries. The results show that the proposed method is robust in finding desired song automatically with hummed query as input.KeywordsContext driftMusic information retrieval (MIR)Perceptual featuresQuery by humming (QBH)Transfer entropy (TE)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call