I've done about 4 years of research on sonification, which is using non-speech audio to represent patterns in data (http://mags.acm.org/interactions/20120102/?pg=37#pg37 for some specifics). This article is a subset of sonification in some ways, since we're representing some quantitative data using auditory parameters.

There's an entire class of scenarios where conventional HCIs can't represent data for analysis: where people have an occupied visual sense (doctors during surgery), where people are mobile, where people are overloaded by visual data (stock analysts), where the visual sense isn't suited for extracting data from noise (during the Voyager 2 mission), etc. We tend to rely only on our visual sense for communicating data, and as we start using computers for data display in more places, we're reaching the limitations of conventional HCI.

My research was on proving the viability of sonification - looking at the accuracy of comprehension, the cognitive and physiological processes, demonstrating shared mental processes with visual graph comprehension, etc. It's still something I'd love to revisit and commercialize someday.