ANALYSIS AND SYNTHESIS

Activities

The Analysis/Synthesis Team conducts research and development activities based on the analysis, tranformation, and synthesis of sound signals. The analysis of sounds includes methods allowing the extraction or automatic structuring of various kinds of signal information, such as the fundamental frequency, or spectral evolutions which determine the pitch and the timbre of the perceived sound.

Information which is not strictly musical is also taken into account, and interests sectors such as industrial acoustics, sound design, and multimedia (for example, automatic indexing of sound recordings). The methods used rely on signal processing, statistical analysis, information theory, learning techniques, and form recognition, but also on knowledge about sound perception and sound production by acoustic systems.

The techniques of transformation and sound synthesis are first conceived to respond to the needs of musicians for the creation of new sounds and new musics. A typical example is the synthesis of a virtual choir by a computer on an opera stage. This work also finds numerous applications in sectors such as mobile telephones, video games, navigation devices, and virtual reality in general.

Analysis and synthesis rely partly on the development of signal models (modelling the effects of sounds produced in terms of signals), and partly on physical models (acoustic modelling of production causes in terms of sound sources). These models are implanted, in the form of software for PC or Macintosh, complete with graphical interfaces specifically developed for professional users and non-professional users alike - musicians, but also sound engineers, acousticians, and music lovers.

Areas of Expertise

Signal processing, statistics, information theory, learning techniques, form recognition.

Themes and Associated Projects

Signal models (Additive, Super VP, Sinola, F0, Carterpillar, Alignement, Séparation), Sound characterisation (Descripteurs, CUIDADO, ECRINS, Indexation), Acoustic system models (Instruments, Inversion, Estimation), Synthesis control and musical applications (Chorales, Phase, Geste).

Team Members

 Head of research : Xavier Rodet
 Engineers and researchers : Joseph Escribe, Jean-Philippe Lambert, Marc LoCascio, Geoffroy Peeters, Axel Roebel
 Guest Researcher : Miroslav Zivanovic
 Doctoral students : André Almeida, Bertrand Delezoide, Wim d'Haes, Arié Livshin, Diemo Schwarz, Kasper Souren, Emmanuel Vincent

Collaborations

ENST (Paris), INRIA (Rennes), Ecole des Mines de Paris, CEA-Saclay, France-Télécom R&D, INA-GRM, Université de Nantes, INPG-Grenoble, Berkeley University (California), San Jose University (California), UPF-Barcelona, Creamware, Artspages, Digigram, Hebrew University (Jerusalem), Université d'Anvers, University of Alberta in Edmonton.