Webinar: Machine Learning Techniques for Head-Related Transfer Function Individualization: A Comprehensive Survey (video)
"A Survey on Machine Learning Techniques for Head-Related Transfer Function Individualization" presented by Mr. Davide Fantini, based on the IEEE Xplore® article with the same title. Published: IEEE Open Journal of Signal Processing, January 2025. Visit the article here.
Immersive spatial audio is essential for realism in applications ranging from virtual reality to teleconferencing. Central to this experience are Head-Related Transfer Functions (HRTFs), which capture how an individual’s anatomy filters sound, enabling its spatialized rendering in a 3D virtual environment.
However, using generic HRTFs often leads to poor localization and unrealistic audio. While acoustic HRTF measurement provides the most accurate personalization, it remains impractical for mass adoption, paving the way for data-driven machine learning solutions.
This webinar presents a comprehensive survey of machine learning techniques for HRTF individualization. We will examine the complete machine learning workflow, systematically categorizing existing methods by the data they employ, such as anthropometric measurements and ear images, preprocessing steps, and model architectures ranging from linear regression to deep neural networks.
Furthermore, the presentation will address critical evaluation methods, challenges in validation, and the current lack of standardization in the field. Attendees will gain valuable insights into the state of the art of data-driven spatial audio, the limitations of current benchmarks, and promising avenues for future research in creating personalized auditory experiences.
