Machine Learning

, Volume 65, Issue 2–3, pp 485–515

Using duration models to reduce fragmentation in audio segmentation

  • Samer Abdallah
  • Mark Sandler
  • Christophe Rhodes
  • Michael Casey
Article

DOI: 10.1007/s10994-006-0586-4

Cite this article as:
Abdallah, S., Sandler, M., Rhodes, C. et al. Mach Learn (2006) 65: 485. doi:10.1007/s10994-006-0586-4

Abstract

We investigate explicit segment duration models in addressing the problem of fragmentation in musical audio segmentation. The resulting probabilistic models are optimised using Markov Chain Monte Carlo methods; in particular, we introduce a modification to Wolff’s algorithm to make it applicable to a segment classification model with an arbitrary duration prior. We apply this to a collection of pop songs, and show experimentally that the generated segmentations suffer much less from fragmentation than those produced by segmentation algorithms based on clustering, and are closer to an expert listener’s annotations, as evaluated by two different performance measures.

Keywords

Segmentation Duration prior MCMC Gibbs sampling Wolff algorithm 

Copyright information

© Springer Science + Business Media, LLC 2006

Authors and Affiliations

  • Samer Abdallah
    • 1
  • Mark Sandler
    • 1
  • Christophe Rhodes
    • 2
  • Michael Casey
    • 2
  1. 1.Queen Mary, University of LondonLondon
  2. 2.Goldsmiths CollegeUniversity of LondonLondon

Personalised recommendations