One of the areas of our application is to make outbound calls that plays a recording to the person called. How can we tell the application to play it if it is a live person and hold until the tone if it is an answering machine?

First, you must be using NMS hardware, as this is the only board we have this support on at present. <br> <br>You must then set/use the following fields in your Place Call step when placing outbound calls: <br> <br>Variable to receive initial voice length: Populates a TeleFlow variable with the number of milliseconds of voice when the call was answered. You then use a Compare step to determine if it was likely a person or machine. (People tend to answer with "Hello", as opposed to "Hello, we're not home right now, etc") We have found that voice lengths less than 3000 milliseconds (3 seconds) are usually real people answering the phone. <br> <br>Silence timeout: The number of milliseconds of silence to wait through after someone answers the phone. We have found 1250 works fairly well for this. (This timeout is what the software uses to determine when the answering side has finished "talking", whether it was a person or machine) <br> <br>Maximum voice length(ms): The maximum milliseconds of voice (talking/audio) to wait through before leaving the Place Call step. <br> <br>So, when the Place Call step succeeds, you can use a Compare to determine whether or not the answering party was a real person, or a machine, based on the "Hello" or "Hello, we're not home etc" scenario above. This isn't perfect, for obvious reasons, but does generally work. Usually, if we think the system was an answering machine, we also play the complete message twice (maybe even three times), just to be sure we left our message.