Show simple item record

dc.contributor.authorSafari, Pooyan
dc.contributor.authorGhahabi, Omid
dc.contributor.authorHernando Pericás, Francisco Javier
dc.contributor.otherUniversitat Politècnica de Catalunya. Departament de Teoria del Senyal i Comunicacions
dc.identifier.citationSafari, P., Ghahabi, O., Hernando, J. Feature classification by means of Deep Belief Networks for speaker recognition. A: European Signal Processing Conference. "23rd European Signal Processing Conference (EUSIPCO) took place 31 August - 4 September 2015 in Nice, France". Niza: Institute of Electrical and Electronics Engineers (IEEE), 2015, p. 2162-2166.
dc.description.abstractIn this paper, we propose to discriminatively model target and impostor spectral features using Deep Belief Networks (DBNs) for speaker recognition. In the feature level, the number of impostor samples is considerably large compared to previous works based on i-vectors. Therefore, those i-vector based impostor selection algorithms are not computationally practical. On the other hand, the number of samples for each target speaker is different from one speaker to another which makes the training process more difficult. In this work, we take advantage of DBN unsupervised learning to train a global model, which will be referred to as Universal DBN (UDBN). Then we adapt this UDBN to the data of each target speaker. The evaluation is performed on the core test condition of the NIST SRE 2006 database and it is shown that the proposed architecture achieves more than 8% relative improvement in comparison to the conventional Multilayer Perceptron (MLP).
dc.format.extent5 p.
dc.publisherInstitute of Electrical and Electronics Engineers (IEEE)
dc.subjectÀrees temàtiques de la UPC::Enginyeria de la telecomunicació::Processament del senyal::Processament de la parla i del senyal acústic
dc.subject.lcshAutomatic speech recognition
dc.subject.otherSpeaker recognition
dc.subject.otherDeep Belief Network
dc.subject.otherRestricted Boltzmann Machine
dc.subject.otherFeature classification
dc.titleFeature classification by means of Deep Belief Networks for speaker recognition
dc.typeConference report
dc.subject.lemacReconeixement automàtic de la parla
dc.contributor.groupUniversitat Politècnica de Catalunya. VEU - Grup de Tractament de la Parla
dc.description.peerreviewedPeer Reviewed
dc.rights.accessRestricted access - publisher's policy
dc.description.versionPostprint (published version)
upcommons.citation.authorSafari, P.; Ghahabi, O.; Hernando, J.
upcommons.citation.contributorEuropean Signal Processing Conference
upcommons.citation.publicationName23rd European Signal Processing Conference (EUSIPCO) took place 31 August - 4 September 2015 in Nice, France

Files in this item


This item appears in the following Collection(s)

Show simple item record

All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder