The SRI NIST 2010 speaker recognition evaluation system

The SRI speaker recognition system for the 2010 NIST speaker recognition evaluation (SRE) incorporates multiple subsystems with a variety of features and modeling techniques. We describe our strategy for this year's evaluation, from the use of speech recognition and speech segmentation to the i...

Full description

Saved in:
Bibliographic Details
Main Authors: Scheffer, Nicolas, Ferrer, Luciana, Graciarena, Martin, Kajarekar, Sachin, Shriberg, Elizabeth, Stolcke, Andreas
Format: Conference Proceeding
Language:English
Subjects:
high-level modeling
Indexes
Mixers
prosody
speaker recognition
system fusion
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The SRI speaker recognition system for the 2010 NIST speaker recognition evaluation (SRE) incorporates multiple subsystems with a variety of features and modeling techniques. We describe our strategy for this year's evaluation, from the use of speech recognition and speech segmentation to the individual system descriptions as well as the final combination. Our results show that under most conditions, the cepstral systems tend to perform the best, but that other, non-cepstral systems have the most complementarity. The combination of several subsystems with the use of adequate side information gives a 35% improvement on the standard telephone condition. We also show that a constrained cepstral system based on nasal syllables tends to be more robust to vocal effort variabilities.
ISSN:1520-6149
2379-190X
DOI:10.1109/ICASSP.2011.5947552