This directory contains results for 8-component MAR with our sustained phone database in:
$ISIP_DATA/research/nonlinear/mix_ar/mar/exp_001/

The speaker recognition result is in:

http://www.ece.msstate.edu/research/isip/projects/nsf_nonlinear/doc/baseline_results/speaker_recognition/mix_ar/mar/exp_001/det_phones_08.jpg

Note the low false alarm probability at the EER point.

As for EM convergence:

http://www.ece.msstate.edu/research/isip/projects/nsf_nonlinear/doc/baseline_results/speaker_recognition/mix_ar/mar/exp_001/em_mc.jpg
for speaker 4516 does not converge well

while for speaker mp,
http://www.ece.msstate.edu/research/isip/projects/nsf_nonlinear/doc/baseline_results/speaker_recognition/mix_ar/mar/exp_001/em_mp.jpg
reasonable convergence is achieved.