I am using Speech Recognition and Body tracking parallely using kinect, while kinect monitors the tracking, speech recognizer takes some words of speech as input automatically without any utterances from the user and interrupts the tracking, please explain
how to avoid this.
Speech recognition can be CPU intensive, and it might be keeping your tracking code from running on a timely basis. And recognition depends on analyzing incoming sounds, so you pay for it whether or not anything is recognized as a command.
I'd start by making sure that I'm doing the two tasks on separate threads, so I could use multiple cores where available.