A set of models dedicated to voice descriptions, with special focus on speaker traits (timbral and non-timbral characteristics, etc)