Paper Info
Title
Robust speaker segmentation for meetings: the ICSI-SRI spring 2005 diarization system
Abstract
In this paper we describe the ICSI-SRI entry in the Rich Transcription 2005 Spring Meeting Recognition Evaluation. The current system is based on the ICSI-SRI clustering system for Broadcast News (BN), with extra modules to process the different meetings tasks in which we participated. Our base system uses agglomerative clustering with a modified Bayesian Information Criterion (BIC) measure to determine when to stop merging clusters and to decide which pairs of clusters to merge. This approach does not require any pre-trained models, thus increasing robustness and simplifying the port from BN to the meetings domain. For the meetings domain, we have added several features to our baseline clustering system, including a “purification” module that tries to keep the clusters acoustically homogeneous throughout the clustering process, and a delay&sum beamforming algorithm which enhances signal quality for the multiple distant microphones (MDM) sub-task. In post-evaluation work we further improved the delay&sum algorithm, experimented with a new speech/non-speech detector and proposed a new system for the lecture room environment.
Year
DOI
Venue
2005
10.1007/11677482_34
MLMI
Keywords
Field
DocType
clustering process,icsi-sri spring,meetings domain,new system,icsi-sri clustering system,robust speaker segmentation,icsi-sri entry,agglomerative clustering,different meetings task,clusters acoustically homogeneous,current system,diarization system,base system
Hierarchical clustering,Broadcasting,Bayesian information criterion,Segmentation,Computer science,Robustness (computer science),Speech recognition,Speaker diarisation,Cluster analysis,Detector
Conference
Volume
ISSN
ISBN
3869
0302-9743
3-540-32549-2
Citations 
PageRank 
References 
34
3.67
4
Authors
4
Name
Order
Citations
PageRank
Xavier Anguera162454.28
Chuck Wooters240458.49
Barbara Peskin317618.45
Mateu Aguiló4343.67