Faculty Host: Professor Ivan W Selesnick
Acoustics may seem a natural field of application for new sampling strategies referred to as "compressed sensing": equations are linear, and acoustic imaging usually requires a large number of measurements. However, it is in most cases not straightforward to describe in which space the signals are (even approximately) sparse, and measurements are not easily designed to satisfy the required incoherence properties.
In this presentation I will describe some experimental work, at the crossroads between wave physics and signal processing, where the compressed sensing framework is successfully applied to a number of acoustics problems, in the audible range: near-field acoustic holography, acoustic source joint localization and characterization, and measurements of the so-called plenacoustic function in room acoustics. In particular, these problems highlight some of the practical difficulties one is faced with: calibration of microphone arrays, computational limitations, and discretization issues.
About the Speaker
Laurent Daudet is professor of physics at Paris Diderot University, France. After a physics education at the Ecole Normale Superieure in Paris, he received in 2000 a Ph.D. degree in applied mathematics from the Universite de Provence. He was then a EU Marie Curie Post-doctoral Fellow at the Centre for Digital Music at Queen Mary University of London, UK. Between 2002 and 2009, he has been working as assistant/associate professor at UPMC - Paris 6, in the D'Alembert Institute for Mechanical Engineering. Since 2009, he is full Professor at the Paris Diderot University - Paris 7, with research at the Langevin Institute for Waves and Images. In october 2010, he was nominated junior member of the Institut Universitaire de France, a 5-year fellowship to foster excellence in academic research. He serves as associate editor for the IEEE Transactions on Audio Speech and Language Processing, and is author or coauthor of more than 100 publications on various aspects of digital signal processing, mostly based on sparse decompositions with applications to audio and acoustics.
Home Page : http://www.institut-langevin.espci.fr/Laurent-Daudet,207