How to improve accuracy

[Carol]

Hi, I'm developing a Speech Application in C#, which accept wav files as
input to create transcript according to what said in the audio file. The code
runs however the results is incorrect , it seems the engine is mostly
guessing and have low accuracy. I'm using dictation grammar right now .
Could any body help me out and told me how to improve the accuracy of wav
file input? I don't know if it is helpful to use the training wizard, since I
feel to train the engine with my voice would not help the engine to recognize
different wav files.

Thanks