This example demonstrates a traditional iVector system evaluated on NIST
   SRE 2016.  It is based on the recipe in ../../sre10/v1/.  In addition to the
   standard features of the SRE10 recipe, it also demonstrates the use of data
   augmentation for PLDA training.
  
   The recipe uses the following data for system development.  This is in
   addition to the NIST SRE 2016 dataset used for evaluation (see ../README.txt).
   
       Corpus              LDC Catalog No.
       SWBD2 Phase 1       LDC98S75
       SWBD2 Phase 2       LDC99S79
       SWBD2 Phase 3       LDC2002S06
       SWBD Cellular 1     LDC2001S13
       SWBD Cellular 2     LDC2004S07
       SRE2004             LDC2006S44
       SRE2005 Train       LDC2011S01
       SRE2005 Test        LDC2011S04
       SRE2006 Train       LDC2011S09
       SRE2006 Test 1      LDC2011S10
       SRE2006 Test 2      LDC2012S01
       SRE2008 Train       LDC2011S05
       SRE2008 Test        LDC2011S08
       SRE2010 Eval        LDC2017S06
       Mixer 6             LDC2013S03
  
   The following datasets are used in data augmentation.
  
       MUSAN               http://www.openslr.org/17
       RIR_NOISES          http://www.openslr.org/28