Speaker segmentation using parallel fusion between three classifiers
Siham Ouamour, Halim Sayoud, Mhania Guerti · 2009
In this paper, we deal with the problem of speaker segmentation. This speciality consists in splitting the audio document into homogeneous areas. Each area is attributed to one speaker. Speaker segmentation (or speaker change detection) consists in detecting the points where the speaker identity changes, in a multi-speaker audio stream. These points or times are called ¿Break Points¿.