Is this Asterisk Application possible?

Question

I have no experience with Asterisk and am looking to replicate an application I have running on the Voxeo platform. The workflow is this:

• Make an outbound call.
• If an answering service is detected, leave a message and then hang up.
• If a human is detected, play some hold music to the human.
• Create another outbound call to a call center.
• When the call center picks up, play a wav file and some text to speech.
• When the call center presses a dtmf key, join the 2 calls together.

Is it possible to replicate this application on an Asterisk solution? What is the difficulty level of putting such a system together. Do you see any pitfalls or areas that would make such a solution problematic to build or run.

Hoping one of you Asterisk geniuses can give me some guidance. Thanks.

Answer 1

Your points 1, 3 and 5 are relatively simple.

Cepstral have some good, cheap voices for tts that asterisk will use.

Point 2 however...

Answerphone detection is a bit of a black art.

We use asterisk to deliver alarm calls. The most reliable scheme we could come up with was asking the user to press a key before we played the message.

If we got a key press, it is a human. We can play the message, and ask for acknowledgement of the alarm.

If we didn't get a key press, we try and detect the end of the answerphone message, and then simply play the message and hangup.

We'd love to be able to do away with the key press, but then run into all sorts of problems trying to get 'detectsilence' to consistently work with the various answerphone messages out there.

Points 4 and 6 I have no experience with.

Answer 2

You can give a try by installing asterisk on Virtual machine.

I have good experience with combination of Sun VM and Elastix BOX, It is easy to start system and begin research of Asterisk goodies.

I am bit a skeptic about detection of answer (human or answering system) Later today when I get more free time Ill try your scenario and try to offer more useful answer.