Is there any code already developed for these issues?

I think use -cutsilence is OK for long input wav files, is that proved to be the best solution or is there any alternatives?

When I get the original output from Julius, it is something like following:

stack empty, search terminate now
0 sentences have found
sentence1: <s>
38800 samples (2.42 sec.)
                    
pass1_best: <s> DIAL NINE NINE
pass1_best_score: -6733.250977

stack empty, search terminate now
0 sentences have found
sentence1: <s> DIAL NINE NINE
11800 samples (0.74 sec.)

How can I decode them to extract the text I want? (I mean the simple speaking String of the audio like I dialed nine one one)

Thank you.

--- (Edited on 8/23/2011 11:07 am [GMT-0500] by qitong) ---

--- (Edited on 8/23/2011 11:38 am [GMT-0500] by qitong) ---