Professor, Department of Engineering
Reykjavik University
I research speech signal processing and speech technology: how the voice source can be estimated from recorded speech, what the signal reveals about a speaker's cognitive and mental state, and how recognition and synthesis can be built for languages with limited resources — Icelandic in particular.
Analysis of the speech signal and of the voice source itself: estimating how the glottis opens and closes, separating source from filter, and characterising voice quality.
Using acoustic and voice-source features to study how cognitive load and mental health conditions are reflected in speech.
Recognition and synthesis methods that work for languages with limited data and resources, with Icelandic as the primary case.
A talk on the Sign and Talk project at Queensland University of Technology, well received by collaborators, plus a note on a QUT student exchange session.
A research visit to Janet Wiles' lab at UQ, conversations on underresourced-language technology, and a case for speech as a signal in behaviour monitoring.
This site replaces the previous static page and now carries research, teaching, activities and publications in one place.