Modelling Individual Acoustics for plausible extended realities

The growing field of extended reality (XR) has shown how technology can expand the physical world with real-time digital information and interactions through applications spanning from hybrid meetings to assisted hearing. While plausible visual information is easily delivered, when not adapted to each user, generic sound rendering lacks realism, increases cognitive load and limits accessibility. Still, a recent study found that only 97% of analysed applications ignore personalised sound rendering due to its complex acoustic requirements.MIA (Modelling Individual Acoustics for plausible extended realities) aims at individualising audio rendering by developing a widely adoptable solution that leverages advancements in psychoacoustics and explainable artificial intelligence (AI). The proposed research effort will (1) explore and determine requirements across multiple acoustic contexts, (2) prototype an explainable AI model to infer individual audio characteristics, and (3) implement a personalisation procedure based on user actions to guarantee plausibility over specific contexts. The resulting software solution aims to provide a toolkit with minimal hardware requirements to integrate plausible audio in XR applications. MIA will foster open-science practices while exploring protection measures for EU product development, supporting the EU's goals of advancing technological innovation and enhancing digital inclusion, particularly in remote work, education, and hearing assistance technologies. To succeed, MIA integrates the diverse fields of expertise of the researcher on auditory modelling and advanced Bayesian statistics, of the host institution on audio-signal processing and XR applications, and of the secondment host on the development of explainable AI. The researcher will, therefore, have the chance to collaborate with multiple leading experts to expand his expertise and professional networks, supporting his development for an independent career.