Human social interactions rely on multisensory cues. In this regard, visual and auditory cues are paramount during the initiation of an interaction. In this preliminary work, we propose an approach to let Intelligent Virtual Agents (IVAs) simulating sound perception capabilities. Our model targets to control IVAs' reactive behaviour through their analysis of perceived other agents' emitted sounds. For that, we explored auditory features close to the human system. CCS CONCEPTS • Computing methodologies → Simulation environments; Motion processing; Procedural animation.