Interactive Voice Response System
State-of-the-art interactive voice (IVR) systems can be implemented with voice
recognition and voice prompts generated using document pages marked up
with the Voice Extensible Markup Language (VoiceXML). Figure 19.6 shows
the call flow example for IVR service.
The service starts with an IVR exchange to determine the wishes of the caller.
The controller, therefore, first proxies the call to the IVR server, so the caller can
interact directly with the IVR server. As in the previous example, the initial
INVITEmessage (1) from the caller has the Request-URI pointing to the con-
troller for this particular service.
After the establishment of the media stream, the IVR will generate a voice
prompt to the caller, along the line of “Welcome to our
speak your ID.” The answer from the caller is transformed from speech to text
and returned in Message 6 of Figure 19.6, HTTPGET, to the controller. The next
VoiceXML script is sent from the controller in the HTTP200 OK(Message 7) to
further prompt the caller for information regarding his or her request. After
the IVR process comes to an end, the last message HTTP200 OK(Message 9)
carries an empty VoiceXML script. The call to the IVR is terminated with a BYE
(Message 10), and the call is forwarded to some other destination with the
INVITEin Message 12.
Interactive voice response systems based on VoiceXML technology can sup-
port several features for voice services:
■■ Text to speech (synthesized speech)
■■ Output of audio files
■■ Voice recognition
■■ DTMF input
■■ Recording of spoken inputSIP Component Services 333