Article ID: | iaor20032385 |
Country: | Belgium |
Volume: | 41 |
Issue: | 1/2 |
Start Page Number: | 69 |
End Page Number: | 81 |
Publication Date: | Jan 2001 |
Journal: | Belgian Journal of Operations Research, Statistics and Computer Science |
Authors: | Pietquin Olivier, Couvreur Laurent, Couvreur Pierre |
Keywords: | statistics: experiment, statistics: multivariate |
In this paper, we introduce an algorithm dedicated to speaker-based segmentation of audio material. The algorithm consists in two distinct procedures, namely splitting and merging. Its performance is assessed on broadcast news recordings provided by the British Broadcasting Corporation. Results show that the splitting is performed with high accuracy and low missed detection rate while the merging procedure provides satisfying results.