Aurora Evaluations

Mission:

Our goal is to evaluate and compare the robustness of feature extraction algorithms in the presence of background noise. These evaluations are being conducted under the Aurora Distributed Speech Recognition working group of The European Telecommunications Standards Institute (ETSI).

What's New:

(11/12/01) Multiple-CPU Eval Package (v1.4.3): Added a utility to remove silence. Fixed a minor bug in the command line interface for lm_scale.
(10/31/01) Short Training Set Definition (v1.4.1): We have included file lists that define the two 7,138 utterance training sets.

Overview: Review the status of the Aurora evaluations, and learn more about the history of this project.

Downloads: Download the latest versions of our recognition software and evaluation scripts.

Publications: Access background information related to this project.

Performance: A quick look at the overall performance of the baseline system using a variety of robust front ends.