Hybrid Multi-Feature Indexing for Music Data Retrieval

Abstract

The management of large collections of music data in a multimedia database has received much attention in the past few years. In the most of current works, the researchers extract the features, such as melodies, rhythms and chords, from the music data and develop indices for helping to retrieve the relevant music efficiently. However, there is only a small number of existing approaches introduced multi-feature index structures for music queries while most of researches are for developing single feature indices. The existing music multi-feature index structures are memory consuming and lack of scalability. In this paper, we will propose a hybrid index structure which can save lots of memory for music multi-feature indexing. Our experimental results also show that the new approach outperforms existing multi-feature index scheme for memory needed.

Stereoscopic Image Feature Indexing Based on Hybrid Grid Multiple Suffix Tree and Hierarchical Clustering

“…The index based on data partition usually has balanced hierarchical trees, such as B-tree, B+-tree, and R-tree formed from B-tree extended to space [15]. The improved tree structures also have R * -tree, X-tree, SS-tree, SR-tree and so on based on classic R-tree [16], and classic suffix tree [17], [18]. The disadvantage of the structures is the problem of overlap while dividing node region boundaries with rectangular or circular.…”

“…In the grid K-medoids clustering, the number of grids is N 2 and the feature points are mapped into each grid according to the values [12]. The clustering is executed in every grid separately and each grid is divided into 9 cells, which is 9 sub-clusters and the value of K is 9 [17]. There are several advantages of the clustering method.…”

“…It has been applied and developed in many algorithms, and it can achieve string matching and query effectively. In this paper, according to the principles of suffix tree and double suffix tree [17], the improved multiple suffix tree is constructed, in which the location area of feature point in the grid is tagged [18]. The grids in row and column are represented as a 0 ,a 1 ,a 2 ,.…”

Real-valued feature indexing for music databases

“…Recently, as the rapid progress in digital representations of music data, how to efficiently manage music data is getting more attentions. There are more and more investigations increasingly attractive in retrieving the music collections such as the Query by Rhythm by Chen, et al [3], Query by Music Segments by Chen, et al [4], Multi-Feature Index Structures by Lee, et al [14] and Lo, et al [21] [22], NonTrivial Repeating Pattern Discovering for music data by Liu, et al [16] and Lo, et al [23], Melodic Matching Techniques by Uitdenbogerd, et al [30], Approximate Melody Matching by Liu, et al [17] [33], Key Melody Extraction and N-note Indexing by Tseng [29], Numeric Indexing for Music Data by Lo, et al [18] [19], and more in [1][10] [15] [18] [20][23] [26] [28].…”

Approximate Searching for Music Data in Real-Valued Feature Indexing1

“…Recently, as the rapid progress in digital representations of music data, how to efficiently manage music data is getting more attentions. There are more and more investigations increasingly attractive in retrieving the music collections such as the Query by Rhythm by Chen, et al [3], Query by Music Segments by Chen, et al [4], Multi-Feature Index Structures by Lee, et al [11] and Lo, et al [17] [18], Non-Trivial Repeating Pattern Discovering for music data by Liu, et al [13] and Lo, et al [19], Melodic Matching Techniques by Uitdenbogerd, et al [25], Approximate Melody Matching by Liu, et al [14], Key Melody Extraction and N-note Indexing by Tseng [24], Numeric Indexing for Music Data by Lo, et al [15] [16], and more in [1] [8] [12][15] [22] [23].…”