Trinity College Dublin

Post-Doc, Centre for Language and Communication Studies

Other Affiliations: Dublin Institute of Technology, Digital Media Centre

Lecturer/Researcher

Digital Media Centre

Thesis Title: Natural emotional speech corpora with large scale emotional dimension ratings.

About

Emotional speech research is of vital importance to future technologies, both in entertainment and ubiquitous computing. While a lot of research has been carried out in the area of emotional speech research, more work needs to be done regarding the type of speech collected, how emotion is elicited and how emotional content is described and annotated.

My PhD was concerned with using dimensional models of emotion (specifically activation and valence) to rate the emotional content of natural emotional speech using large-scale online listening tests. The natural speech is derived from task-based success-failure Mood Induction Procedures (MIPs) and aims to capture the important communicative underlying emotional states.

The speech was then rated using a web based tool created in Adobe Flex 3 with the ratings being written back to the database for analysis.

In doing this, I addressed a number of issues: examining and determining the validity of two-dimensional models, the relationship of certain acoustic parameters to the valence dimension, as well as the activation dimension and the use of social-networks as a means of obtaining large number of listeners for the rating of emotional speech.