Evaluation of HRTF Models for Binaural Cue Adaptation

Abstract:
Binaural recordings with artificial heads or ear-mounted microphones are an efficient way to capture spatial audio scenes. The idea of Binaural Cue Adaptation (BCA) is to enable head-tracked reproduction of such recordings by transforming binaural signals recorded with a certain head orientation to signals appropriate for a listener with a different head orientation. It uses parametric filtering based on a signal model with Head-Related Transfer Functions (HRTFs), which are estimated using HRTF models and Direction of Arrival (DOA) estimation. This contribution investigates the effect of a mismatch between the HRTF models used by the algorithm and the HRTFs embedded in a recording. A generic HRTF model is proposed for the case when the recording head is unknown, which yields better results than the previously used spherical head model.