In the context of IRISA D5 (Signal, Image & Robotics) department’s scientific seminars, we held a thematic seminar on localization, separation and tracking of audio sources, on February, 19th and 20th 2013.
Audio source separation consists in extracting individual contributions of different audio sources (speakers, musical instruments…) to a mixture. It is now mature enough to envision applicative transfer. Related research issues are not all solved yet, especially in real-world context : reverberant environments, diffuse or moving sources… Source separation is deeply related to the problems of locating and tracking sources, which are generally a front-end to source separation systems.
This two-day workshop aimed at giving a tutorial overview of source localization and separation scientific background, as well as advanced presentations on some recent approaches to these problems in different contexts.
The seminar was open to academic and industrial experts, as well as Ph.D candidates, postdoctorate researchers and colleagues from neighboring research fields who wish to learn the basics of source separation, or improve their knowledge of the domain and of its current theoretical and practical challenges. The program is labeled « complément scientifique » by Ecole Doctorale Matisse.
Detailed program can be downloaded here: JTD52013
You can download here the speakers’ slideshows :
- Emmanuel Vincent: source localizatio jtd5_evincent1, source separation jtd5_evincent2, source classification jtd5_evincent3;
- Gilles Chardon: Vekua approximations and applications to soundfield sampling and source localization jtd5_gchardon
- Laurent Daudet: Sampling plenacoustic fields jtd5_ldaudet
- Nancy Bertin: Some sparse and cosparse tools and methods for sound source localization and separation jtd5_nbertin