Summary and initial results of the 2013-2014 speaker recognition i-vector machine learning challenge

Bansé, Désiré; Doddington, George R.; Garcia-Romero, Daniel; Godfrey, John J.; Greenberg, Craig S.; Martin, Alvin F.; McCree, Alan; Przybocki, Mark; Reynolds, Douglas A.

doi:10.21437/Interspeech.2014-86

iBet uBet web content aggregator. Adding the entire web to your favor.

Link to original content: https://doi.org/10.21437/Interspeech.2014-86

Summary and initial results of the 2013-2014 speaker recognition i-vector machine learning challenge

Désiré Bansé, George R. Doddington, Daniel Garcia-Romero, John J. Godfrey, Craig S. Greenberg, Alvin F. Martin, Alan McCree, Mark Przybocki, Douglas A. Reynolds

During late-2013 through early-2014 NIST coordinated a special i-vector challenge based on data used in previous NIST Speaker Recognition Evaluations (SREs). Unlike evaluations in the SRE series, the i-vector challenge was run entirely online and used fixed-length feature vectors projected into a low-dimensional space (i-vectors) rather than audio recordings. These changes made the challenge more readily accessible, especially to participants from outside the audio processing field. Compared to the 2012 SRE, the i-vector challenge saw an increase in the number of participants by nearly a factor of two, and a two orders of magnitude increase in the number of systems submitted for evaluation. Initial results indicate the leading system achieved an approximate 37% improvement relative to the baseline system.