ATVS-UAM NIST SRE 2010 SYSTEM

This paper describes the system submitted by ATVS-U AM to the 2010 edition of NIST Speaker Recognition Evaluat ion (SRE). Instead of focusing on multiple, complex and heavy systems, our submission is based on a fast, light a nd efficient single system. Sample development results with Engl ish SRE08 data (data used in the previous evaluation in 2008) are 0.53% EER (Equal Error Rate) in tel-tel (telephone d ata used for training and testing) male data (optimistic eva luation), going up to 3.5% (tel-tel) and 5.1% EER (tel-mic, t elephone data for training and microphone data for testing) in pessimistic cross-validation experiments. These res ults are achieved with an extremely light system in computat ional resources, running 77 times faster than real time.

[1]  Hynek Hermansky,et al.  RASTA processing of speech , 1994, IEEE Trans. Speech Audio Process..

[2]  Sridha Sridharan,et al.  Feature warping for robust speaker verification , 2001, Odyssey.

[3]  Sridha Sridharan,et al.  Explicit modelling of session variability for speaker verification , 2008, Comput. Speech Lang..

[4]  Patrick Kenny,et al.  Support vector machines versus fast scoring in the low-dimensional total variability space for speaker verification , 2009, INTERSPEECH.

[5]  Patrick Kenny,et al.  Eigenvoice modeling with sparse training data , 2005, IEEE Transactions on Speech and Audio Processing.

[6]  Douglas A. Reynolds,et al.  Speaker Verification Using Adapted Gaussian Mixture Models , 2000, Digit. Signal Process..

[7]  Douglas A. Reynolds,et al.  The SuperSID project: exploiting high-level information for high-accuracy speaker recognition , 2003, 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03)..

[8]  Executive Summary.  EVALUATION PLAN , 2014 .

[9]  Douglas A. Reynolds,et al.  Channel robust speaker verification via feature mapping , 2003, 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03)..