[Freeswitch-users] Voice pattern detection

I am trying to detect if a caller is an automated greeting voice. And if so, take an action. I have samples of the caller recording that I am looking to match. So this is like a really complex tone detection I guess. It would work like this. - Call comes in - We answer/bridge the call - We start to listen for about 10 seconds. - During this time we are trying to match a snippet of a sound sample (say 2 or 3 seconds worth) we have recorded on a file on the server. We are trying to match this sound sample to the caller side only. - If we hear a match, we take the action. Or if we don't hear the match, we might take a different action. Any of this sound doable? Any guidance on how to accomplish this? -Frank