IITG-Indigo System for NIST 2016 SRE Challenge

Nagendra Kumar, Rohan Kumar Das, Sarfaraz Jelil, Dhanush B.K., H. Kashyap, K. Sri Rama Murty, Sriram Ganapathy, Rohit Sinha, S.R. Mahadeva Prasanna

This paper describes the speaker verification (SV) system submitted to the NIST 2016 speaker recognition evaluation (SRE) challenge by Indian Institute of Technology Guwahati (IITG) under the fixed training condition task. Various SV systems are developed following the idea-level collaboration with two other Indian institutions. Unlike the previous SREs, this time the focus was on developing SV system using non-target language speech data and a small amount unlabeled data from target language/ dialects. For addressing these novel challenges, we tried exploring the fusion of systems created using different features, data conditioning, and classifiers. On NIST 2016 SRE evaluation data, the presented fused system resulted in actual detection cost function ( actDCF) and equal error rate ( EER) of 0.81 and 12.91%, respectively. Post-evaluation, we explored a recently proposed pairwise support vector machine classifier and applied adaptive S-norm to the decision scores before fusion. With these changes, the final system achieves the actDCF and EER of 0.67 and 11.63%, respectively.

 DOI: 10.21437/Interspeech.2017-1307

Cite as: Kumar, N., Das, R.K., Jelil, S., B.K., D., Kashyap, H., Murty, K.S.R., Ganapathy, S., Sinha, R., Prasanna, S.M. (2017) IITG-Indigo System for NIST 2016 SRE Challenge. Proc. Interspeech 2017, 2859-2863, DOI: 10.21437/Interspeech.2017-1307.

  author={Nagendra Kumar and Rohan Kumar Das and Sarfaraz Jelil and Dhanush B.K. and H. Kashyap and K. Sri Rama Murty and Sriram Ganapathy and Rohit Sinha and S.R. Mahadeva Prasanna},
  title={IITG-Indigo System for NIST 2016 SRE Challenge},
  booktitle={Proc. Interspeech 2017},