Sorry for the confusion..
Yes, initially it was about null hypothesis. That is solved now.
Later, it is the issue of slow recognition. Sometimes it takes as long as even 2-3 minutes to return a hypothesis.(Which leads to memory warning followed by the app crashing).
So I replaced my language array in the sample app and tried. Recognition is slow there too.. So I asked if it was a matter of the size of the vocabulary..
As you suggested, I will try varying the vadThreshold and the secondsOfSilence to see any changes in the performance.
Also, I have removed the ‘suspendRecognition’ call now.
Thank you. I will keep you posted.