Embodiments herein provide an augmented reality (AR) system that uses sound localization to identify sounds that may be of interest to a user and generates an audio description of the source of the sound as well as AR content that can be magnified and displayed to the user. In one embodiment, an AR device captures images that have the source of the sound within their field of view. Using machine learning (ML) techniques, the AR device can identify the object creating the sound (i.e., the sound source). A description of the sound source and its actions can outputted to the user. In parallel, the AR device can also generate AR content for the sound source. For example, the AR device can magnify the sound source to a size that is viewable to the user and create AR content that is then superimposed onto a display.