Application des machines à vecteurs support mono-classe à l'indexation en locuteurs de documents audio
Belkacem FERGANI, Manuel DAVY, Amrane HOUACINE.

This paper addresses a new approach based on the kernel change detection algorithm introduced recently by Desobry et al. This new algorithm is applied to the speaker change detection and clustering tasks, which are the key issues in any audio indexing process. We show the efficiency of the method through several experiments using RT'03S NIST data. We discuss also the parameters tuning and compare the results to the well known GLR-BIC algorithm.