[June 26, 2002]

Plum VoiceXML IVR Enhanced With Open Source Festival TTS Components

The Plum Group announced the integration of the Festival text-to-speech engine into the Plum VoiceXML interactive voice response (IVR) system. The open source Festival TTS system, developed by the Center for Speech Technology Research (CSTR) at the University of Edinburgh, is now available as a low cost speech module for the Plum Voice system.

"Using open source technology for key speech modules in our system gives us the flexibility to fine tune the technology -- and the pricing -- to best match the needs of our clients. These technologies are not new. What's new is that now, with VoiceXML, we have a framework to easily put these technologies to work," said Andrew Kuan, VP of Technology.

VoiceXML is a World Wide Web Consortium (W3C) standard for specifying platform and technology independent voice applications. It allows developers to leverage existing web and intranet back-end infrastructure to support new voice services. "Just as HTML and the Internet are replacing traditional, proprietary client-server systems, VoiceXML is replacing proprietary systems for IVR applications," said Kuan.

Originally developed at the University of Edinburgh, Festival has now been downloaded more than 50000 times and is used by hundreds of speech synthesis researchers around the world both in academia and in industry. Using the FestVox tools from Carnegie-Mellon University, voices have been built for Festival for over 40 languages. This list includes: English (various dialects) French, German, Spanish (various dialects), Italian, Swedish, Scots, Welsh and Irish Gaelic, Basque, Catalan, Breton, Korean, Japanese, Chinese, Croatian, Haitian Creole, Telugu, Hindi, Nepali, Turkish, Greek and even Klingon. Some are made available for free, others are commercial.

Festival also supports SSML, an SGML based markup standard for tagging text with prosody and emphasis cues.

In addition to the Festival TTS engine, Plum also offers AT&T Natural Voices speech synthesis as an option. "Licensing fees for proprietary commercial speech modules are often a large expense for IVR systems. Festival gives our customers a free, open source alternative," said Kuan.

[ Back To TMCnet.com's Homepage ]