HRTF Upsampling

Machine-learning methods and international benchmarking for efficient personalised HRTF measurement

Personalised head-related transfer functions (HRTFs) can improve spatial localisation and immersion, but obtaining a sufficiently dense set of individual measurements is time-consuming and requires specialist facilities. My research develops machine-learning methods for reconstructing high-spatial-resolution HRTFs from only a small number of measured directions, making personalised spatial audio more practical and scalable.

This research has progressed from generative adversarial networks operating on spatial and spherical-harmonic representations, through denoising and autoencoder-based approaches, to spatially aware transformer architectures. The work considers not only spectral reconstruction accuracy, but also spatial consistency and perceptually relevant evaluation.

I also played a central role in organising the Listener Acoustic Personalisation Challenge 2024 (LAP24), serving as Chair of Website and Dissemination and contributing to the design, delivery and evaluation of its HRTF-upsampling task. LAP24 established an international benchmark through which algorithmic and machine-learning approaches could be compared using common datasets, metrics and perceptual evaluation.

Selected publications

  • HRTFformer: A Spatially-Aware Transformer for Individual HRTF Upsampling in Immersive Audio Rendering
    IEEE Transactions on Multimedia, 2026.
    Paper

  • Listener Acoustic Personalisation Challenge LAP24: Perceptual Evaluation of HRTF Upsampling
    EAA Forum Acusticum, European Congress on Acoustics, 2026.

  • A Machine Learning Approach for Denoising and Upsampling HRTFs
    European Signal Processing Conference (EUSIPCO), 2025.
    Paper

  • Listener Acoustic Personalisation Challenge – LAP24: Head-Related Transfer Function Upsampling
    IEEE Open Journal of Signal Processing, 2025.
    Paper · Challenge · Code

  • Head-Related Transfer Function Upsampling Using an Autoencoder-Based Generative Adversarial Network with Evaluation Framework
    Journal of the Audio Engineering Society, 2025.
    Paper

  • HRTF Upsampling With a Generative Adversarial Network Using a Gnomonic Equiangular Projection
    IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024.
    Paper

  • HRTF Spatial Upsampling in the Spherical Harmonics Domain Employing a Generative Adversarial Network
    International Conference on Digital Audio Effects (DAFx), 2024.
    Paper

View related publications on the VIABAL website