Events Calendar

Classification of Speech Under Cognitive Load

Speaker: Julien Epps, The University of New South Wales, Sydney,School of Electrical Engineering and Telecommunications

Talk Title: Classification of Speech Under Cognitive Load

Abstract: The estimation of a user’s cognitive load has traditionally involved manual and/or post-hoc approaches such as subjective self-rating, performance measures and reaction time. Recently there has been a movement to automate cognitive load estimation, towards indicators that are sensitive to real time fluctuations in mental load during tasks. Speech is a natural signal to consider, being relatively free of the intrusiveness, inconvenience or privacy concerns associated with other signals such as video, EEG or GSR. Little is known about the effects of cognitive load on speech, however. In this presentation, results from recent investigations into vocal source-related and vocal tract-related parameters will be discussed, and promising system configurations for cognitive load classification will be presented. Finally, a brief overview of related current work on mental state recognition from speech and other methods for cognitive load estimation will be given.

Biography: Julien Epps is a Senior Lecturer with The University of New South Wales School of Electrical Engineering and Telecommunications. He has a joint position as a Senior Researcher with National ICT Australia (a government research lab), as part of a longer-term project investigating real time, non-intrusive cognitive load measurement using behavioural and physiological signals. In recent years he has worked mainly on the recognition of emotion, depression and cognitive load from speech and on speaker recognition. For example, as part of a funded project based at UNSW, he is investigating methods for dealing with speaker and phonetic variability in the emotion recognition problem.

Host: Prof. Shrikanth Narayanan

Location: RTH 320

Audiences: Everyone Is Invited

Contact: Mary Francis