First of all, let me say I'm an Arduino newbie; I currently don't even own one. I've been looking for a way to automate voice calls and found the Arduino GSM shield.
Basically, what I want to do is collect some data (this part should be no problem with the Arduino from what I've read so far) and read back this data when somebody calls the Arduino. Both having voice files for the possible data and using a text-to-speech library such as Cantarino would be feasible.
I'm now wondering whether that would be possible with an Arduino Uno and a GSM shield, because the overview states that
Through the modem, it is possible to make voice calls. In order to speak to and hear the other party, you will need to add a speaker and microphone.
So is there a way to somehow get audio input from software instead of hardware?
Integration between Cantarino and any modem doesn't exist currently. You can do some tests connecting microphone input of GSM Shield to some input according to Cantarino.
It should be possible with some additional circuitry.
Have a look at the wave shield: Overview | Wave Shield | Adafruit Learning System
There should be an explanation how that shield works somewhere in there. Basically it uses reads a wav file form an SD card and plays it using a digital to analog converter. You could set up a similar thing, but instead of outputting the audio to a speaker, you output the audio to the microphone input of the gsm shield or something like that.
I'm sure with some tinkering you could figure this out
I was looking to do the same thing with preprogrammed responses based on OpenCV Events. The idea was to follow an object, identify it, categorized it, then either text or call my cell. For calling, it would have predetermined responses to speak to the other end. I do t own any shields, but I do have the OSEPP Uno R3+.