Geração de ambientes sonoros imersivos e não-lineares a partir de exemplos monofónicos

Abstract

The creation of a soundscape can have some complexity, especially if we are working with an audio recording, this can limit sound designer actions during post-production. The most common problem these experts face is the duration of the recorded file that can be difficult to extend. Currently, there are no sound designing tools that allow for automatic segmentation and classification of the content of the audio file and, as a consequence, change the duration of the soundscape. This dissertation presents the development of a prototype for a digital audio workstation that receives an audio file and outputs in real-time an automatic background generation, segmented foreground events into clusters, spatialization of said foreground events, and its placement in the background timeline.