Message Fidelity as a Measure of Speech Intelligibility

Prof. Bastiaan Kleijn (Victoria University of Wellington)

APPLIED SIGNAL PROCESSING SERIES

DATE: 2013-12-05
TIME: 10:00:00 - 11:00:00
LOCATION: RSISE Seminar Room, ground floor, building 115, cnr. North and Daley Roads, ANU
CONTACT: JavaScript must be enabled to display this email address.

ABSTRACT:
We introduce the mean probability of correctly decoding a message as a measure of intelligibility. Existing measures of intelligibility can be interpreted as simplifications of this measure. A natural new simplification suggests the use of the similarity in spectral dynamics of the noisy speech and the noise-free speech as a practical means to measure intelligibility. We use this approach for the enhancement of speech in a noisy environment and for the selection of a particular formulation out of a set of formulations for improved clarity in text to speech. The method has low computational cost and requires little delay, thus facilitating practical real-time implementation. Experimental results confirm the effectiveness of the approach.
BIO:
Bastiaan joined Victoria in 2010. He is also a Professor (part-time) at Delft University of Technology in the Multimedia Signal Processing Group. Until his Ph.D. students are finished he also remains a Professor at KTH - the Royal Institute of Technology - in Stockholm, where he was Head of the Sound and Image Processing Laboratory before moving to New Zealand. He was a founder of Global IP Solutions, a company that developed audio and video processing technologies for communication over the Internet for companies such as Skype, IBM, Cisco, and Google; it was acquired by Google in 2010. Before going to Sweden, Bastiaan worked in the Research Division of Bell Laboratories in Murray Hill, New Jersey, mostly in the area of speech processing. He holds a Ph.D. in Electrical Engineering from Delft University of Technology, a Ph.D. in Soil Science from the University of California (UCR), an MS in Physics (UCR) and an MS in Electrical Engineering from Stanford. He is an IEEE Fellow.