Rescaling clustering trees using impact ratios for robust hierarchical speaker clustering
Data(s) |
03/12/2014
|
---|---|
Resumo |
We present a novel method for improving hierarchical speaker clustering in the tasks of speaker diarization and speaker linking. In hierarchical clustering, a tree can be formed that demonstrates various levels of clustering. We propose a ratio that expresses the impact of each cluster on the formation of this tree and use this to rescale cluster scores. This provides score normalisation based on the impact of each cluster. We use a state-of-the-art speaker diarization and linking system across the SAIVT-BNEWS corpus to show that our proposed impact ratio can provide a relative improvement of 16% in diarization error rate (DER). |
Formato |
application/pdf |
Identificador | |
Publicador |
The New Zealand Institute of Language, Brain and Behaviour (NZILBB) |
Relação |
http://eprints.qut.edu.au/75968/1/paper28_final_SST2014.pdf http://www.nzilbb.canterbury.ac.nz/graphics/SST%202014_Proceedings%20optimized.pdf Ghaemmaghami, Houman, Dean, David, Kalantari, Shahram, & Sridharan, Sridha (2014) Rescaling clustering trees using impact ratios for robust hierarchical speaker clustering. In Proceedings of the 15th Australasian International Conference on Speech Science and Technology (SST 2014), The New Zealand Institute of Language, Brain and Behaviour (NZILBB), Christchurch, New Zealand, pp. 159-162. http://purl.org/au-research/grants/ARC/LP130100110 |
Direitos |
Copyright 2014 Please consult the authors |
Fonte |
School of Electrical Engineering & Computer Science; Institute for Future Environments; Science & Engineering Faculty |
Palavras-Chave | #090609 Signal Processing #speaker diarization #speaker clustering #cluster impact ratio |
Tipo |
Conference Paper |