Postdoc Position: Audio Signals Segmentation and Indexing, Laboratoire Traitement et Communication de l’Information, Paris, France

The LTCI lab (Laboratoire Traitement et Communication de l’Information) which is a joined research lab between CNRS and GET/TIcom Paris is proposing a postdoctoral position in Audio signals segmentation and indexing, to start in September/October 2006.

Project Description: The focus of this project is on audio indexing and on content-based information retrieval especially for radiophonic audio streams. For such streams, the audio signal gathers on a single track (or file) numerous events or combination of events (speech, music, applause, environmental noise, jingles, etc) that are important to automatically detect. In fact, it is known that efficient speech/music segmentation leads to improved performances for speech recognition or speaker tracking. However, beyond the speech/music segmentation, it is also important to consider more complex situations (speech detection on musical background, solo detection in a music performance, singing voice segments localisation, genre or orchestration estimation etc.). Hence, one of the main objectives of this project is to obtain an automatic segmentation of the different types of segments (speech, music) including mixed segments in developing new statistical approaches for novelty detection and content structuring, in developing new methods for speech enhancement (or singing voice enhancement) with musical background (and vice versa for musical sources identification) and in developing new methods for audio information extraction (automatic extraction of main melody, harmony, rhythm and genre) from musical signals. This research work will fit in the framework of several national and international collaborative projects and in the first place in the European network of excellence IST-Kspace that aims at building an open and expandable framework for collaborative research in semantic inference for semi-automatic annotation and retrieval of multimedia content.

Candidate Profile: As minimum requirements, the candidate will have:

• A PhD in audio signal processing, speech processing, statistics, machine learning, computer science, electrical engineering, or a related discipline.
• Familiarity with audio signal processing
• Programming skills The ideal candidate would also have:
• Experience with corpus-based methods.
• Solid experience of research work materialized by publications in conferences or/and journals
• Experience with machine learning and excitement about interdisciplinary work.
• Autonomy and excitement to work in a team
• Some musical experience

Other Information
Preferred starting date: September or October 2006
Location: LTCI / TIcom Paris, 37 rue Dareau, 75014 Paris, FRANCE
Duration : 12 months

Competitive salary
The LTCI lab is located in the heart of Paris (France) one of the culturally most exciting, diverse, and inclusive cities in the world (Web sites: Signal and Image Processing department, http://www.tsi.enst.fr ; GET/TIcom Paris: http://www.enst.fr ; LTCI (in French): http://www.ltci.enst.fr/ )

For more information, please contact :
Prof. Gael RICHARD (Gael.Richard at enst.fr / +33 1 45 81 73 65)
Prof. Yves GRENIER (Yves.Grenier at enst.fr)
Prof. Henri MAITRE (Henri.Maitre at enst.fr)