Paper Info
Title
Speaker diarization in meeting audio
Abstract
This paper describes speaker diarization system on a NIST Rich Transcription 2007 (RT-07) Meeting Recognition evaluation data set for the task of Multiple Distant Microphone (MDM). Our implementation includes three components: initial clustering, non-speech removal and cluster purification. Initial clusters are generated using Directional of Arrival (DOA) information and bootstrap clustering. Multiple GMM modeling for speech/non-speech classification is employed for non-speech removal component. In addition, a novel system fusion strategy using information from Receiver Operating Curve (ROC) is proposed for non-speech removal component. Finally, consensus clustering approach together with iterative GMM clustering method is employed for speaker cluster purification. The system achieves the overall DER of 10.81%.
Year
DOI
Venue
2009
10.1109/ICASSP.2009.4960523
ICASSP
Keywords
Field
DocType
multiple gmm modeling,cluster purification,meeting audio,non-speech removal,speaker diarization system,bootstrap clustering,novel system fusion strategy,non-speech classification,iterative gmm clustering method,non-speech removal component,initial clustering,speaker recognition,decision support systems,direction of arrival,receiver operating curve,speech,data mining,machine learning,receiver operator curve,probability density function,natural languages,speaker diarization,modeling,adaptive filters,erbium,sun,speech processing,tin
Speech processing,Pattern recognition,Computer science,Speech recognition,Speaker recognition,Consensus clustering,NIST,Artificial intelligence,Speaker diarisation,Cluster analysis,Bootstrapping (electronics),Microphone
Conference
ISSN
Citations 
PageRank 
1520-6149
5
0.52
References 
Authors
7
4
Name
Order
Citations
PageRank
Tin Lay Nwe147934.59
Hanwu Sun29814.15
Haizhou Li33678334.61
Susanto Rahardja4652102.05