George Erhart

Research Scientist


I am currently working on two projects that explore different aspects of using SIP based Interactive Voice Response (IVR) systems. The first project is called Context and Communications Integration (CCI) and involves the integration of SIP into a contact center as a replacement for dedicated Computer Telephony Integration servers that are found in traditional contact centers. SIP allows for significantly richer communications between endpoints in the contact center and can be used to push transaction data with the call thus removing the need to consult 3rd party servers for the information. Additionally, the presence services of a SIP infrastructure can be used as the clearing house for resource availability events within the call center. Finally, traditional contact center features are limited to the control of the telephony switch that runs the contact center. The distributed nature of SIP allows the contact center to blur the institutional boundaries to expand contact center capabilities to areas that might not have been included before and to make multiple, geographically separated contact centers inter-operate seamlessly.

The second project involves the incorporation of video into an automated VoiceXML based dialog system. The traditional role of  the IVR system is to execute VoiceXML scripts to prompt callers for information and pass entered data to contact center applications servers for processing. With the deployment of new video capable 3G cell phones and video capable broadband phones, video is added as an additional means of communicating information to the caller. The project is exploring how to use a video channel to enhance the standard voice dialog to improve the accuracy and usability of self service systems.

My past projects include the creation and integration of a VoiceXML 2.0 compliant browser for the Avaya IVR platform. I also have maintained and active role in the VoiceXML Forum VXML compliance committee. I have also worked on various speech recognition focused projects including client/server speech recognition architectures, large vocabulary name dialing, and speaker trainable name dialing.


I have published one article in the 1996 ICSLP proceedings entitled A User Configurable System for Voice Label Recognition, co-authored with R.C. Rose, E. Lleida, and R.V. Grubbe.