logo Ircam

Sample Orchestrator 2

Activity Period: 
01/11/2010 - 30/10/2013

This suite of the Sample Orchestrator project aims to build new functions targeting a new generation of software samplers, integrating; three complementary aspects:

Modeling and expressive synthesis of musical instrument sounds, real time sound using hybrid approaches to sound spatialization, real time orchestration. The analysis/synthesis team works on the first topic.

Instrument Models for Expressive Sound Synthesis

Today’s sound transformation algorithms allow signal transpositions of more than a factor 2 without introducing artifacts. A problem is the fact that the transformed signals do not merge acoustically with the untransformed sounds of the same instrument. This creates severe constraints for the use of signal transformation algorithms in sampling based applications. To be able to solve this problem instrument models that describe the timbre modifications that are related to pitch and energy changes or expressive modulations are required. Few results exist that allow describing the instrument dependent transformations that have to be applied during transposition of individual notes. Recent advances in physical modeling of the wave propagation in the trumpet have shown that the nonlinear effects accompanying level changes can be modeled with a high degree of precision by means of a simple structure of filters and instantaneous nonlinear operations (T. Hélie, D. Roze, 2008a)(T. Hélie, B. Laroche, 2008b). These models, however, are not available for all instruments and the question of adapting these models to particular instruments has not yet been addressed. The idea of the master and PhD thesis of Henrik Hahn (H. Hahn, A. Roebel, J. Burred, S. Weinzierl, 2010a) (H. Hahn, A. Röbel, 2012a) and of the instrument models developed in the Sample Orchestrator II project is to derive a parametrized source filter model of the complete timbre space of a musical instrument that allows the production of sound transformations that can be used as expressive sound gestures in sampling based synthesis applications.

Project details


ANR (Agence Nationale de la Recherche)



SOR2 Project page




UniverSons, CNSMDP (Conservatoire national supérieur de musique et de danse de Paris).

Principal Research Theme: Signal Analysis, Sound Synthesis and Treatment
Specific Research Theme: Sinusoidal model
Associated Project: Sample Orchestrator 2
Associated Software: Pm2, SuperVP
Project Finance: ANR
Associated People: Axel Roebel, Henrik Hahn