Statistical methods have played a significant role in making speech recognition successful in a diverse range of applications including interactive voice response systems and hands-free control in an automobile. Central to this success has been our ability to statistically optimize every aspect of the system using closed-loop data-driven training techniques. We are now developing a book that attempts to provide a perspective on these techniques from a unified point of view.

As with all things in our organization, we are committed to open source distribution. Hence, you will be able to find the most current copy of the manuscript on our web site at Statistical Optimization in Speech Recognition. We are looking for people interested in reviewing preliminary copies of the book chapters and providing feedback. If you want to be involved, please send email to isip_help@ece.msstate.edu and ask to be included in our distribution list. If you just want to randomly comment on the manuscript, send your comments to isip_help@ece.msstate.edu as well. Expect the first release, which will include an outline and some introductory notes, to be available by March 1.