dc.contributor |
Universitat Politècnica de Catalunya. Departament de Teoria del Senyal i Comunicacions |
dc.contributor |
Universitat Politècnica de Catalunya. VEU - Grup de Tractament de la Parla |
dc.contributor.author |
Hernando Pericás, Francisco Javier |
dc.date |
2012 |
dc.identifier.citation |
Hernando, J. On the use of agglomerative and spectral clustering in speaker diarization of meetings. A: The Speaker and Language Recognition Workshop. "Odyssey 2012: The Speaker and Language Recognition Workshop". Singapur: 2012, p. 130-137. |
dc.identifier.uri |
http://hdl.handle.net/2117/18147 |
dc.language.iso |
eng |
dc.rights |
info:eu-repo/semantics/openAccess |
dc.subject |
Àrees temàtiques de la UPC::Enginyeria de la telecomunicació::Processament del senyal::Processament de la parla i del senyal acústic |
dc.subject |
Automatic speech recognition |
dc.subject |
Reconeixement automàtic de la parla |
dc.title |
On the use of agglomerative and spectral clustering in speaker diarization of meetings |
dc.type |
info:eu-repo/semantics/publishedVersion |
dc.type |
info:eu-repo/semantics/conferenceObject |
dc.description.abstract |
In this paper, we present a clustering algorithm for speaker
diarization based on spectral clustering. State-of-the-art diariza-
tion systems are based on agglomerative hierarchical clustering
using Bayesian Information Criterion and other statistical met-
rics among clusters which results in a high computational cost
and in a time demanding approach. Our proposal avoids the use
of such metrics applying Euclidean distances on the eigenvec-
tors computed from the normalized graph Laplacian. A hybrid
system is proposed in which HMM/GMM modelling and Viterbi
alignment are still applied, but the BIC for merging and stop-
ping criterion are substituted by a spectral clustering algorithm.
Once an initial segmentation is obtained and the clustering align-
ment is computed using the Viterbi algorithm, the remaining
clusters are modeled by stacking the means of the Gaussians in
a super vector. In such a space single value decomposition of
the associated normalized graph Laplacian is computed. Most
similar clusters are merged based on the Euclidean distances
in resulting eigenspace. Cluster number estimation is based on
analyzing eigenstructure of the similarity matrix by selecting
a threshold on the eigenvalues gap. In experiments, this ap-
proach has obtained a comparable performance to the traditional
AHC+BIC approach on the Rich Transcription conference eval-
uation data. Although it still relies on Gaussian modelling of
clusters and Viterbi alignment, the proposed approach leads to a
system which runs several times faster than traditional one. |
dc.description.abstract |
Peer Reviewed |