Grouping Songs Using String-Matching and Lloyd's Algorithm

Abstract

In this project, we study algorithmic solutions to cluster songs based on their rhymes and temporal similarity. Existing commercial software evaluates two songs’ similarities based on their metadata. Instead, we use a value to denote a song’s temporal pattern so that two songs having similar rhymes are with similar values. The algorithmic technique that we are using is the k-mean algorithm, which is to group songs into k groups and each group has all its songs similar to each other. Our experimental results show that our approach is promising and such grouping results also give us hints in characterizing music.