Using OpenVXI version 3.4, VoiceGlue can process VoiceXML 2.0 code and implements many 2.1 features.

Current Voiceglue Features

• VXML 2.0 interpreter with some VXML 2.1 features
• Flite TTS engine
• Pre-recorded audio playback in Asterisk audio formats
• Asterisk DTMF detection
• SRGS DTMF grammars with no more than one variable length part
• Recording in ulaw (audio/basic) format
• Shared caching of downloaded audio files
• Shared caching of TTS prompts
• Shared caching of SRGS DTMF grammars
• High-performance, high-scalability architecture
• Distributable processing components

Planned Voiceglue Features

• Speech recognition using LumenVox engine
• All VXML 2.1 features
• Additional audio formats for record and playback
• Higher-quality TTS package integration

Because of Asterisk’s great flexibility, VoiceGlue’s implementation of VoiceXML is not restricted to SIP telephony, but can use the entire range of Asterisk supported telephony channels.

A particular feature of the VoiceGlue implementation is the decomposed architecture of the solution. The VoiceXML interpreter (including OpenVXI) can be run on a separate host from Asterisk. In fact, that is how it is currently being run in one production site. This allows easy multi-host VoiceXML interpreter configurations and associated scalability without affecting the scalability of Asterisk. Of course, the VoiceXML interpreter could also reside on the same host as Asterisk in cases where this was desired.