Analysis/synthesis and Spatialization of Noisy Environmental Sounds

The use of stochastic modeling is discussed for analysis/synthesis and transformation of environmental sounds. The method leads to perceptually relevant synthetic sounds based on the analysis of natural sounds. Applications are presented, such as sound effects using parametric signal transformations, or data compression. Moreover, we propose a method which efficiently combines the stochastic modeling with 3D audio techniques. This architecture offers an efficient control of the source width rendering that is often an important attribute of noisy environmental sounds. This control is of great interest for virtual reality applications to create immersive 3D scenes.