FOR IMMEDIATE RELEASE

ejTalk and FlexVoice Powered Demos

October 11, 2001 -- ejTalk announced today that we are empowering some of our web-based demonstrations with a small footprint TTS from a member of the Mindmaker group: FlexVoice. This is one of the steps in our plan to put in place the components and methods that open up these technologies to the next tier of consumer.

Be sure to check it out......Digital Shakespeare!

ejTalk has long held that our long term goals should be to create interactive, distributed and evolutionary development methodologies that result in truly natural synthetic conversational interfaces.

The FlexVoice product appeals to us because it has a hybrid approach that promises to allow more manipulation of the prosody ("how" we say things) than most of the large footprint concatenated approaches. One of the trade-offs in TTS has been larger concatenated systems that have a more human-like tone (they are just lots of very short recordings that are reassembled) versus tiny footprint formant systems that compute the vocal tract as if it were a flexing organ pipe. The resultant tones are always recognizably not human, but they can generate a range of prosody that is essentially boundless. The formant-based systems can be made to speak like a human but they don't sound human.

We look forward to exercising the FlexVoice technology in the web as well as the host based environment.

For more information contact ejTalk.