Speaker Recognition from Coded Speech Using Support Vector Machines

Artur Janicki , Tomasz Staroszczyk


We proposed to use support vector machines (SVMs) to recognize speakers from signal transcoded with different speech codecs. Experiments with SVM-based text-independent speaker classification using a linear GMM supervector kernel were presented for six different codecs and uncoded speech. Both matched (the same codec for creating speaker models and for testing) and mismatched conditions were investigated. SVMs proved to provide high accuracy of speaker recognition, however requiring higher number of Gaussian mixtures than in the baseline GMM-UBM system. In mismatched conditions the Speex codec was shown to perform best for creating robust speaker models.
Author Artur Janicki (FEIT / IT)
Artur Janicki,,
- The Institute of Telecommunications
, Tomasz Staroszczyk - [Warsaw University of Technology (PW), MNiSW [80]]
Tomasz Staroszczyk,,
- Politechnika Warszawska
Publication size in sheets0.5
Book Habernal Ivan, Matoušek Václav (eds.): Text, Speech and Dialogue, Lecture Notes In Computer Science, no. 6836, 2011, Springer Berlin Heidelberg, ISBN 978-3-642-23537-5, 1-443 p., DOI:10.1007/978-3-642-23538-2
bfm%3A978-3-642-23538-2%2F1.pdf / 166.39 KB / No licence information
Keywords in Englishspeaker recognition, speaker classification, speech coding, support vector machines. Artificial Intelligence (incl. Robotics), Database Management, Data Mining and Knowledge Discovery, Information Storage and Retrieval, Information Systems Applications (incl.Internet), speaker classification, speaker recognition, speech coding, support vector machines, User Interfaces and Human Computer Interaction
URL http://link.springer.com/chapter/10.1007/978-3-642-23538-2_37
Languageen angielski
chp%3A10.1007%2F978-3-642-23538-2_37.pdf 212 KB
Score (nominal)13
Score sourcejournalList
Publication indicators WoS Citations = 13; Scopus Citations = 11; GS Citations = 27.0
Citation count*25 (2020-03-25)
Share Share

Get link to the record

* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.
Are you sure?