Speech-driven phone applications are the result of technological developments in computer telephony, speech technology (recognition & synthesis), and web technology. VoiceXML is the standard language in which most current speech-driven phone applications are implemented.
Live VoiceXML applications are based on the following architectural components:
- A VoiceXML platform (a.k.a. browser), with integrated speech technology and telephony components: this is where live applications physically run
- An application server: this can be a classic web server, or a Voice Application Management System (e.g. VoiceObjects, Audium or VocaBase)
- A VoiceXML application: is hosted on the application server, where it is retrieved by the VoiceXML platform
- Data: they are stored somewhere on a database, and typically accessed through a web server
- A service number: starts with 0800, 070, or 090x (in Belgium); this is the number that end users call. The call is transferred to the VoiceXML platform
The elements above constitute the physical building blocks of an application. An element which is at least as important, but immaterial, is the Voice User Interface (VUI): it implements the interaction between the application and the caller. VUI design is a critical succes factor in any speech-driven phone application.