Assistant Professor, Université de Sherbrooke
Role: Lead for Acoustic Localization
I'm a member of IntRoLab at the Interdisciplinary Institute for Technological Innovation. I am also a member of INTER, CRASH, and CdRV. My current research mainly focuses on robot audition. This is an exciting field, as the end goal is to allow natural human-robot interaction with voice in everyday life environments. This task involves numerous challenges that go beyond traditional far-field speech recognition approaches. In addition to dealing with room reverberation, interfering noise and/or competing speakers, robots must ignore the noise generated by their own actuators, which we refer to as ego-noise, and perform online recognition with little latency. Whilst deep learning-based approaches have been successful in improving speech recognition robustness, they rely on a large amount of data and cloud computing. This is a major challenge in robotics, as the amount of experimental data is limited (each robot is unique!), and autonomous robots must perform computations on board. My current research articulates around four main themes: 1) sound source localization, 2) speech enhancement, 3) sound classification, and 4) ego-noise reduction.
Education
PhD in Electrical Engineering, Université de Sherbrooke, 2017
MS in Electrical Engineering, Université de Sherbrooke, 2011
BS in Electrical Engineering, McGill University, 2009