Have you ever wondered why sound localization is essential to our daily lives? The ability to accurately determine the direction and distance of sounds is critical for our safety and well-being, as listeners rely on it in various situations, such as crossing a busy street. Likewise, it is fundamental when conversing with a friend in a noisy restaurant where spatial information allows the listener's brain to ignore irrelevant sounds.
In Barumerli et al. [1], we demonstrate how a computational model (i.e. a piece of software) can predict this ability to localize a sound source. We used probability theory with Bayes' theorem to represent how the human brain might solve the localization task. As a result, we can predict listener performances in estimating a sound source's horizontal and vertical directions in a static scenario.
Our model is helpful in many ways. For example, its modular architecture can be extended to predict additional behavioural abilities (e.g., estimate a sound source's distance). Or, it can speed up application development by helping to assess the localization impairment introduced by hearing aids, boost public spaces' safety, or improve user experience in virtual reality. We show in Daugintis et al. [2], how the model can help to individualize the audio rendering system to enhance immersion in a virtual reality environment.
In conclusion, sound localization is a fundamental aspect of our daily lives, and our model predicts such ability. The model targets both basic research and applications, and we released its implementation as open source within AMT 1.3.
[1] Barumerli, R., Majdak, P., Geronazzo, M., Meijer, D., Avanzini, F. and Baumgartner, R., (2023) "A Bayesian model for human directional localization of broadband static sound sources" in: Acta Acustica, 7, p.12. DOI: doi.org/10.1051/aacus/2023006
[2] Daugintis, R., Barumerli, R., Picinali, L. and Geronazzo, M., 2023, June. Classifying Non-Individual Head-Related Transfer Functions with A Computational Auditory Model: Calibration And Metrics. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1-5). IEEE.