«

»

Vendredi 11 octobre : Exposé de Mark Plumbey, Director of C4DM, « Making sense of sound and music »


A partir de 14h00

Amphithéâtre – Espace Conférences Inria

Centre de recherche Inria Rennes- Bretagne Atlantique

Campus de Beaulieu

En partenariat avec

titre

Old worn down eighties cassette with band pulled out

Programme :

– 14h00 – 15h00 : Exposé de Mark Plumbley, Professor & Director of the Centre for Digital Music (C4DM), Queen Mary University of London.

« Making sense of sound and music »

Résumé : Sound and music surrounds us all the time. Often we hear all this without really noticing: it just forms part of the background to our lives. The human ability to listen to sounds is something that is very hard for computers; but we are now beginning to build sound processing methods that can help us. In this talk I will discuss some of these techniques that can separate out different sound sources from a mixture, follow the notes and the beats in a piece of music, or show us the sound in new visual ways. These « machine listening » algorithms offer the potential to make sense of the huge amount of sound and music in our digital world: helping us to analyze sounds like heartbeats or birdsong, find the music we want in huge collections of music tracks, or to create music in new ways.

MarkPlumbleyBiographie : Prof. Mark Plumbley is Director of the Centre for Digital Music (C4DM) at Queen Mary University of London. His research interests include the analysis of audio and music signals, including beat tracking, automatic music transcription and source separation, using techniques such as information theory and sparse representations. He is Principal Investigator on several current EPSRC grants, including « Sustainable Software for Digital Music and Audio Research » and a Platform Grant, and he holds an EPSRC Leadership Fellowship. He leads the UK Digital Music Research Network and is a member of the IEEE Audio and Acoustic Signal Processing Technical Committee.

– 15h00 – 16h00 :

Exposé de Sylvain Argentieri, ISIR-UPMC, UMR7222 CNRS.

« Towards active binaural robot audition »

Résumé : Audition is often described as one of the most important human senses, particularly because of its central role in communication and socialization. But surprisingly, the interest of this modality for robotics arose only in the 2000s, brought to evidence by Cognitive robotics and Human-Robot Interaction. In recent years, robot audition research has covered numerous fields, from source localization to computational auditory scene analysis. Following a short review of the specificities of the robotics context, the different contributions on the robot audition research field of the PAM (Multimodal Active Perception) group, part of the ISIR (Institute for Intelligent Systems and Robotics) lab will be introduced. After a summary of our past work on microphone arrays, this talk will focus on binaural approaches, which exploit only two microphones, generally located inside an artificial head. In this complex context, robotics provides a new solution: the use of  binaural sensor movement information. Several active approaches based on this feature will then be introduced: localization of a potentially moving sound source through stochastic filtering algorithms, as well as a new approach to auditory perception relying on a sensory-motor integration inspired by recent works connected to the psychology of perception.

Exposé de Guillaume Gravier, IRISA/CNRS & Inria Rennes – Bretagne Atlantique.

« Linking media contents for navigation : the TEXMIX story »

Résumé : Multimedia content linking is an emerging field of research with ongoing projects (e.g., LinkedTV, inEvent, BBC iPlayer) and numerous potential industrial applications in the field of multimedia asset management (linked digital archives, second screen applications or rich MOOC). At IRISA/INRIA Rennes, we are pursuing research in this line of work, developing TEXMIX, an emblematic demo illustrating the notion of navigation in broadcast news archives where links are created based on content similarities. I will present the demo and brush the technologies inside (e.g., ASR, named entity extraction, topic segmentation, motif discovery, audio summarization) concluding with open issues and research directions.

Télécharger l’affiche du Colloquium du 11 octobre 2013.