Room transfer function estimation from background noise processing

This study investigates how the background noise naturally present in a room can reveal the acoustic properties of the space. We hypothesize that the transfer functions in the room emerge from observing background noise events over time and across different locations. Similar techniques have been widely used in seismology and underwater acoustics to estimate Green's functions, but less so in room acoustics and audio applications. In this work, we develop a model that enables to estimate room transfer functions in the presence of isotropic noise, and validate it experimentally. By analyzing synchronous noise observations from multiple microphones distributed throughout the room, we can approximate the transfer functions between them. The approach provides information about sound propagation between all pairs of sensors subject to the noise field, unlike classical measurement methods that focus on propagation between a given source and all sensors in the field. The principles examined can be of significant value for room acoustics and immersive audio technologies.