Binaural sound spatilization
My current research work at IRCAM with Dr. Olivier WARUSFEL deals with binaural source spatialization. More precisely, I'm focusing on the estimation of individualized Head Related Transfer Functions (HRTFs) using a binaural recording in unsupervised conditions. Because the use of non-individual HRTFs for binaural sound source spatialization can cause perceptual artefacts and the generalization of such measurements is hardly accessible for large public, HRTF individualization methods present an alternative to prevent from measuring HRTFs.
Multichannel source separation
During my PhD, I worked on blind source separation for robot audition under the supervision of Pr. Yves GRENIER and Dr. Karim ABED-MERAÏM. This work was part of the ROMEO project led by Aldebaran Robotic. This project aimed to build an humanoid (ROMEO) that can act as a comprehensive assistant for persons suffering from loss of autonomy. My task in this project was focused on the blind source separation using a microphone array (more than 2 sensors). I developed a two step blind source separation algorithms where the first step consists in a fixed beamforming preprocessing using the Head Related Transfer Functions (HRTF) of the robot and the second step is a blind source separation based on the l1 norm minimization and applied to the outputs of the beamformers.
Then I also worked on the implementation of the developed source separation algorithm on C language and run it in real time on the robot's chipset and the incorporation of the visual modality into an audio source separation algorithm to help the localization and the separation of the speakers.
For more details about the proposed source separation algorithms and the obtained results, please visit my publications page.
During my master thesis, I worked on the study and conception of HMM-SVM hybrid algorithms for environmental sound classification under the supervision of Slim ESSID and it was realized in the Image and Signal Processing (TSI) department of Telecom ParisTech. My mater report is available in french here.