Strong and weak Head-related transfer functions: The eHRTF analytical framework

This article introduces an analytical framework for modeling head-related transfer functions (HRTFs) from a listener-centered perspective. The distinction between strong (or general) HRTFs, aiming for idealized physical acoustic fidelity, and weak (or narrow) HRTFs, prioritizing perceptual adequacy in task-specific contexts, frames the contrast in multiple contrasting definitions and scientific methodologies by drawing inspiration from the debate in artificial intelligence. The proposed formalism adopts a Bayesian structure that models HRTFs through a state-space formulation capturing anatomical, contextual, experiential, and task-related factors: the eHRTF. The “e” emphasizes the egocentric perspective, transforming HRTFs from static measurements into mutable auditory representations continuously updated through the listener's feedback. Satisfaction regions are defined in probabilistic terms and characterize how different classes of HRTFs, i.e., individual, generic, super, and personalized, meet perceptual requirements under varying tasks and their complexity.