Hello,
I am using Sphinx4beta2 for Mandarin speech recognition. I am using my own acostic and languqge models , It's pretty good and works well, however the time of the recognition is a little longer:the total recognition time is about 6 seconds,if the training database is a little smaller,the time seems  a little shorter, I wonder to kown whether the recognition time depends on the size of the training database , How to short the recognition time ?

Thanks for your help and suggestions.