Fast query by humming system based on complex multiscale music entropy and CMMEB Kd tree

摘要

Query by Humming (QBH) Music Information Retrieval (MIR) system is one of the enlivened areas of research for matching hummed query to music excerpts existing within the music database. This paper proposes a fast QBH content based music retrieval technique based on the estimation of Complex Multiscale Music Entropy (CMME). The proposed concept exploits the statistical reliability through the CMME for approximation of music signals. In order to enhance the speed of the system, the CMME feature vectors of music database are indexed through Complex Multiscale Music Entropy Balanced (CMMEB) Kd tree leading to reduced search space and retrieval time. Eventually, CMME feature vectors are extracted from humming query for retrieval of the target song from music database. The experimental results effectually locate the humming queries and state the prominence of CMME and CMMEB Kd tree for retrieving the target song by humming.