Paper Info
Title
The lia-eurecom RT'09 speaker diarization system: Enhancements in speaker modelling and cluster purification
Abstract
There are two approaches to speaker diarization. They are bottom-up and top-down. Our work on top-down systems show that they can deliver competitive results compared to bottom-up systems and that they are extremely computationally efficient, but also that they are particularly prone to poor model initialisation and cluster impurities. In this paper we present enhancements to our state-of-the-art, top-down approach to speaker diarization that deliver improved stability across three different datasets composed of conference meetings from five standard NIST RT evaluations. We report an improved approach to speaker modelling which, despite having greater chances for cluster impurities, delivers a 35% relative improvement in DER for the MDM condition. We also describe new work to incorporate cluster purification into a top-down system which delivers relative improvements of 44% over the baseline system without compromising computational efficiency.
Year
DOI
Venue
2010
10.1109/ICASSP.2010.5495088
ICASSP
Keywords
Field
DocType
pattern clustering,speaker recognition,LIA-Eurecom RT'09 speaker diarization system,NIST RT evaluation,baseline system,bottom-up system,cluster purification,speaker modelling enhancement,top-down system,DER,MDM,SDM,Speaker diarization,cluster purification,speaker clustering,speaker segmentation
Data modeling,Pattern recognition,Pattern clustering,Computer science,Speech recognition,NIST,Speaker recognition,Artificial intelligence,Speaker diarisation,Baseline system,Hidden Markov model,Cluster analysis
Conference
ISSN
Citations 
PageRank 
1520-6149
9
0.74
References 
Authors
5
3
Name
Order
Citations
PageRank
Simon Bozonnet1242.51
nicholas evans259454.41
corinne fredouille353744.53