Date: 13 May 2014
Lecture by Mark Thomas, Microsoft Research.
The desire for hands-free telephony and immersive teleconferencing/telepresence systems has sparked much interest in signal processing for microphone and loudspeaker arrays. Many of the tools that are now commonplace in acoustic arrays have roots in entirely different fields; most beamforming techniques stem from phased-array antenna technology and much of Fourier acoustics is borrowed from quantum theory, both emerging around the beginning of the 20th Century. While the impact of these tools is undeniable, many design procedures nevertheless make assumptions that are reasonable for the original scenarios but are not so valid in the acoustic case. For instance, transducers may be assumed to be omnidirectional, have a flat frequency response, or be matched in sensitivity. Ideally the design of processing algorithms should account for non-ideal behavior by using measured directivity and radiation patterns that can vary significantly in the real world. In this talk we investigate how practical measurements can be made, how array signal processing can benefit from this information in the acoustic beamforming scenario, and other applications in the field of audio and acoustics.
Mark Thomas received the M.Eng. degree in Electrical and Electronic Engineering and the Ph.D. degree from Imperial College London, London, U.K. in 2006 and 2010 respectively. His research interests include signal processing for speech including source modelling and speaker identification, and particularly multichannel acoustic signal processing for beamforming, acoustic echo cancellation, dereverberation, and spatial audio capture and rendering. He is currently a Researcher with Microsoft Research, Redmond, USA. Dr. Thomas is a member of the IEEE Signal Processing Society, the American Acoustical Society and the Audio Engineering Society.