- Voice Application Development Using VoiceXML
- Voice Applications and Portals
- Architecture of a Voice Application
- The VoiceXML Language
- Conclusion
- Additional Resources
Architecture of a Voice Application
As you might expect, a voice application is broken up into several distinct "tiers," much like traditional Web applications. Several unique features set voice apps apart, however. To begin, consider the client: a telephone handset (either landline or wireless). No specific software runs on the client, so the emphasis is on information delivery, not fancy graphics, client-side scripting, or other tricks of the Web trade. When the user's voice hits the telephony server (typically equipped with multiple Dialogic or other manufacturer's voice-resource cards), it must first be processed by special speech-recognition software, available from vendors such as Conversay, Speechworks, and Nuance. In VoiceXML-compliant products (such as the Nuance Voice Web Server), the VoiceXML interpreter is integrated at this stage in order to connect user voice commands with VoiceXML menu options (more on this in the next section). The next stage in the process involves retrieval of datawhether textual, database-driven, or prerecorded audio files stored on a server. Data being delivered back to the user also passes through the telephony server and is then heard over the user's handset. Again, this process allows truly thin clients (dumb voice handsets) to access and make use of Web content alongside their more capable relatives (PDAs, WAP phones, and so on).