期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2016
卷号：4
期号：1
页码：629
DOI：10.15680/IJIRCCE.2016.0401142
出版社：S&S Publications
摘要：Music information retrieval (MIR) is a science of retrieving information from music signal. MIR is a critical and challenging research topic, especially for real - time online search of similar songs over internet. Accurate and compact representation of music signals is a key component of large content - based music information retrieval (CBMIR). Here we work on ho w to index, and quickly and reliably retrieve relevant songs from a large - scale dataset of music audio tra cks according to melody similarity. The proposed system involves compact representation of audio tracks by exploiting music content. It is done by recognizing and extracting chord progressions. The chord progressions are recognized from music signals bas ed on a supervised statistical learning model. The extraction of chord progressions is based on support vector machine (SVM) and Hidden Markov Models (HMM). Here a chord progression histogram (CPH) is computed from each audio track as a mid - level feature, which retains the discriminative capability in describing audio content. Further the efficient organization of audio tracks is done according to their CPHs by using hash table. A set of dominant chord progressions (CPs) of each song is used as the hash key . Chord progressions and key information can serve as a robust mid - level representation and indexing for a variety of MIR tasks. A system is developed to measure the retrieval performance using precision and recall for a given query from a set of relevant music documents
