| Résumé |
This internship report deals with the synthesis of sound textures based on a statistical description. The synthesis model under investigation was proposed by Josh McDermott and Eero Simoncelli in 2011 (McDermott & Simoncelli, 2011). First the definition of what is a sound texture will be discussed. This is followed by a description of model under investigation, insights on the theoretical aspects of the model and on its implementation will be given. The limitations of the algorithm will be discussed and related to the theoretical definition of sound textures. Some proposition to possibly improve the implementation are made. Finally some preliminary work on identifying links between higher-level parameters, such as density, and low-level statistics will be presented. |
