There may be more optimal and efficient methods of human-to-machine auditory interaction than the linear/conversational approach we use in most human-to-human conversation. Yet in 2006 consumer voice-driven applications all emulated this linear approach. We explored a system that leveraged the cocktail party effect [wikipedia] to allow users to “skim” multiple audio streams and quickly pinpoint specific messages.


Detail of MessengerMail flow diagram.

See the demo videos for novice and expert use, or the project website for more information.

Fall 2006, Collaborators: Ray J. Su, Hwi Kyoung Lee, Melissa Ludowise.