Okay, so the military really wants speech to speech capabilities... just think of the possibilities for TTS! Egads, just think of the complexities for TTS and speech recognition, actually. Sombody better let those folks know that it's going to take more than 18 months to get something deployed along those lines. And if they did, it would be highly context specifc. They are talking about having Standard Arabic as the dialect... which might be okay for some folks to understand (let's say that Standard Arabic is the output)... but what about all the potential dialects going into the SR system? Sounds too good to be true. (Heck, what do I know, I only work in the speech industry).