13th Annual SIGdial Meeting on Discourse and Dialogue, 2012
Department of Computer Science, University of Texas at El Paso
|
Abstract: Models of dialog state are important, both scientifically and practically, but today's best build strongly on tradition. This paper presents a new way to identify the important dimensions of dialog state, more bottom-up and empirical than previous approaches. Specifically, we applied Principal Component Analysis to a large number of low-level prosodic features to find the most important dimensions of variation. The top 20 out of 76 dimensions accounted for 81% of the variance, and each of these dimensions clearly related to dialog states and activities, including turn taking, topic structure, grounding, empathy, cognitive processes, attitude and rhetorical structure. |
materials for the extended talk at the 8th Speech Synthesis Workshop, 2013: slides and handout