Efficient kNN search in polyphonic music databases using a lower bounding mechanism

Abstract

Querying polyphonic music from a large data collection is an interesting and challenging topic. Recently, researchers attempt to provide efficient techniques for content-based retrieval in polyphonic music databases where queries can also be polyphonic. However, most of the techniques do not perform the approximate matching well. In this paper, we present a novel method to efficiently retrieve k music works that contain segments most similar to the user query based on the edit distance. A list-based index structure is first constructed using the feature of the polyphony. A set of candidate approximate answers is then generated for the user query. A lower bounding mechanism is proposed to prune these candidates such that the k answers can be obtained efficiently. The efficiency of the proposed method is evaluated by real data set and synthetic data set, reporting significant improvement over existing approaches in the response time yielded.

Continuous query processing over music streams based on approximate matching mechanisms

“…Techniques for string matching can be adopted to enhance the performance of query processing. However, the conventional methods, such as the suffix tree [44] and the n-gram index [17] will generate a complex and costly index for searching owing to the variety of events in polyphonic music [32]. Therefore, these techniques are not suitable for our problem.…”

“…Note that f can also be a negative integer. This approach can also be applied to a broader area of applications by using the contour representation, i.e., the first-order interval, for polyphonic music introduced in [32]. The first-order interval collects the differences in pitch scales between two consecutive events.…”

“…Let λ denote the null character. Three edit operations and the corresponding costs [32] are defined as follows: …”

“…The edit distance [27,42] is commonly used in string matching to measure the difference between two strings. Various applications, such as music retrieval, root-cause analysis, speech recognition, plagiarism detection, and DNA alignment [32,38,43,50], use the edit distance as their similarity measures. For the problem we consider, the edit distance is intuitive for revealing the difference between two musical segments and is simple for users to understand.…”

“…To keep the property of continuity for music segments, we design a new kind of CQ, named the sequence query (SQ), as a sequence of events, where an event denotes the set of notes played at the same time [16,32]. This pitch representation is also named homophonic reduction [46], assuming the independence among the notes with overlapping duration.…”