Light Weight Mobile Device Targeted Speaker Clustering Algorithm

Olli Vuorinen, Tommi Lahti, Satu-Marja Mäkelä, Johannes Peltola

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

2 Citations (Scopus)


In this paper we present a novel light weight speaker clustering algorithm based on the Bayesian Information Criterion (BIC). Algorithm utilises BIC profiles, which were earlier used for False Alarm Compensation (FAC) in our Speaker Change Detector (SCD). Proposed speaker segmentation followed by a light weight clustering is targeted to segment and label mobile device recordings directly in the device itself. Thus the main criterion in algorithm design was to maintain high detection accuracy while keeping computational costs in low level. Clustering algorithm gave F-score performance of 0.90 for speaker segmentation, which is 29% relative improvement compared to baseline [1] results. Speaker segment labelling performance was 88%, when the number of speakers was undetermined. The experimental results indicate that our unsupervised speaker clustering algorithm is sufficiently effective and efficient for speaker segmentation applications in mobile devices.
Original languageEnglish
Title of host publication2008 IEEE 10th Workshop on Multimedia Signal Processing
Place of PublicationAustralia
PublisherIEEE Institute of Electrical and Electronic Engineers
ISBN (Electronic)978-1-4244-2295-1
ISBN (Print)978-1-4244-2294-4
Publication statusPublished - 2008
MoE publication typeA4 Article in a conference publication
Event2008 International Workshop on Multimedia Signal Processing, MMSP 2008 - Cairns, Australia
Duration: 8 Oct 200810 Oct 2008


Conference2008 International Workshop on Multimedia Signal Processing, MMSP 2008
Abbreviated titleMMSP 2008


  • robustness
  • Gaussian distribution
  • mobiel handsets
  • speech
  • clustering algorithms

Fingerprint Dive into the research topics of 'Light Weight Mobile Device Targeted Speaker Clustering Algorithm'. Together they form a unique fingerprint.

Cite this