Smart songs selection in playlists using parallel k-means clustering

Title

Smart songs selection in playlists using parallel k-means clustering

Creator

Manoj P.; Saleema J.S.

Description

Most songs today are of different tempo, pitch and time signature. In a music player application, the typical shuffle picks the succeeding song or preceding song at random with no parameters to choose the songs. Different songs from different genres can have a tempo range anywhere between forty beats per minute and three hundred beats per minute. In this paper, the quick and efficient parallel k means clustering algorithm is implemented in Hadoop on the million-song dataset subset to form clusters for the songs based on tempo and pitch. The aim of this paper is to reduce the variation that occurs when a typical shuffle picks the succeeding song at random. This variation can be in the form of tempo or other parameters. The formation of clusters and intern the reduction in the variation of tempo can be used in a new 'smart shuffle'. After the clusters have been formed, the smart shuffle picks the songs within that specific cluster. This paper aims at reducing the variation by 50%. This would have many musical benefits and would also be more pleasing to the listener. 2018 IAEME Publication.

Source

International Journal of Civil Engineering and Technology, Vol-9, No. 3, pp. 778-788.

Date

2018-01-01

Publisher

IAEME Publication

Subject

Hadoop; Parallel K-means clustering; Pitch; Tempo; Variation

Coverage

Manoj P., Department of Computer Science, Christ Deemed to Be University, Hosur Road, Bhavani Nagar, Bengaluru, Karnataka, 560029, India; Saleema J.S., Department of Computer Science, Christ Deemed to Be University, Hosur Road, Bhavani Nagar, Bengaluru, Karnataka, 560029, India

Rights

Restricted Access

Relation

ISSN: 9766308

Format

Online

Language

English

Type

Article

Identifier

https://www.scopus.com/inward/record.uri?eid=2-s2.0-85044778264&partnerID=40&md5=fdd4c338f9105f1065d0e4d3c5180e86