The MIT LL 2010 speaker recognition evaluation system: Scalable language-independent speaker recognition

Research in the speaker recognition community has continued to address methods of mitigating variational nuisances. Telephone and auxiliary-microphone recorded speech emphasize the need for a robust way of dealing with unwanted variation. The design of recent 2010 NIST-SRE Speaker Recognition Evaluation (SRE) reflects this research emphasis. In this paper, we present the MIT submission applied to the tasks of the 2010 NIST-SRE with two main goals— language-independent scalable modeling and robust nuisance mitigation. For modeling, exclusive use of inner product-based and cepstral systems produced a language-independent computationallyscalable system. For robustness, systems that captured spectral and prosodic information, modeled nuisance subspaces using multiple novel methods, and fused scores of multiple systems were implemented. The performance of the system is presented on a subset of the NIST SRE 2010 core tasks.

• Type: Journal
• Year: 2011
• Where: ICASSP
• Authors: Douglas E. Sturim, William M. Campbell, Najim Dehak, Zahi N. Karam, Alan McCree, Douglas A. Reynolds, Fred Richardson, Pedro A. Torres-Carrasquillo, Stephen Shum
• Keywords: ICASSP 2011, Language-independent Scalable Modeling, Signal Processing, Speaker Recognition, SRE 2010 Core