Speech Recognition Problems

Hello: I am trying to integrate Nuance with Asterisk and have few issues.

Present setup (AGI) plays a message, then records the channel with 2 seconds silence detection and then passes the file to nuance server with grammar etc.

The issue we have here is that the system has to wait till the entire message is played and we cannot prompt it in between. The users have to wait till the entire greeting is over before the system starts recording.

I am looking for ways to record the speech (if any) during playback of a file.

What is the best route?

Also, is there a way to eliminate the noise in the wav file? Nuance would do it but wanted to check if asterisk does it somehow.