12th Annual Conference of the International Speech Communication Association

Florence, Italy
August 27-31. 2011

Language Recognition via i-vectors and Dimensionality Reduction

Najim Dehak (1), Pedro A. Torres-Carrasquillo (2), Douglas Reynolds (2), Reda Dehak (3)

(1) MIT, USA
(2) MIT Lincoln Laboratory, USA
(3) LRDE, France

In this paper, a new language identification system is presented based on the total variability approach previously developed in the field of speaker identification. Various techniques are employed to extract the most salient features in the lower dimensional i-vector space and the system developed results in excellent performance on the 2009 LRE evaluation set without the need for any postprocessing or backend techniques. Additional performance gains are observed when the system is combined with other acoustic systems.

Full Paper

Bibliographic reference.  Dehak, Najim / Torres-Carrasquillo, Pedro A. / Reynolds, Douglas / Dehak, Reda (2011): "Language recognition via i-vectors and dimensionality reduction", In INTERSPEECH-2011, 857-860.