I was having a terrible time calibrating my microphone and Realtek soundcard to give good quality speech recognition. No matter how much training I did, I still got a lot of mistakes compared to DNS 8. Just for fun, I slowed down the process priority (using task manager) from normal to low for the application NatSpeak.exe. Lo and behold, speech recognition and speed increased 100%. Who would have thunk it?
I am using a core two duo, T8100 CPU at full speed to which is 2.1 GHz. Just a wild guess that there is a timing issue during the processing of the sound input and that slowing things down a little bit reduces the hiccups. Anyway, it works for me.
