So many of the new speech based applications that we see fail in the same way that VCR engineering has failed. Of all the owners of VCRs, only a small percentage actually use them to record anything. And of those that do record, most of them will do so by pushing the "record" button as the show begins. An amazingly small number of people actually program the VCR to record at a later time.

All of the functionality is in the VCR, but the method, even though it is just a few button pushes, is too large a cognitive load. It just doesn't resonate with people. (Admittedly, the VCR interface has become much better in recent years.)

The VCR engineers created the functions and provided the methods to do the job, but people hated it because was "hard to remember" and certainly not natural.

Speaking is one of the most natural things we as humans do. We do it quite well. So, why do so many speech interfaces seem as hard to use as a VCR?

Beyond the raw programming necessary to connect to the basic speech technology functionality lies the task of fitting the interaction to a human.

• Prototype -- a tool to discover how they will really use it
• Dialog -- a specification for a set of potential conversational paths
• Grammar -- a linguistic description of what needs to be recognized (and just as importantly what does not)

A VCR still has value (i.e. playing movies) even if you ignore the programmable recording interface but, a voice based application is all about the interface. The voice interface will make the difference between failure or success.

We understand speech technology. We have been there, done that.

ejTalk can:

• explain how people will interact with your application
• work with your technical staff in the early development phases and components
• be an ongoing resource