Visualizing Voice

Audio communication research to date has been primarily dominated by work in the areas of speech recognition, transmission and compression, synthesis, computer music theory, and some music information retrieval. Looking at many research laboratories and universities, we tend to find audio processing groups focusing exclusively on the above areas.

In the area of Human Computer Interaction (HCI), research in audio is in the minority.  For example, there are several textual search engines and even image search engines, yet barely a voice browser for public use.  One reason is that a voice or audio browser relies heavily on speech recognition and audio classification which are not very accurate in general use scenarios. Given different speakers and different speaking environments, the problem becomes increasingly more difficult.

In this talk, we are taking a step back and looking at voice from a simpler perspective.  We will show examples of conversational dynamics, retrieval through the use of a real time voice visualization on a tabletop, and examples of new interactions by using this interface as a social mirror.