Sequential and direct access of head-related transfer functions (HRTFS) for quasi-continuous angular positions

Binaural rendering is based on head-related transfer functions (HRTFs) to reproduce the soundfield around the listener using stereo-headphones. Especially, for the perception of moving sound sources, a high angular resolution of the measured HRTFs is needed. A continuous HRTF representation with quasi-infinite azimuthal resolution can be obtained with a system identification approach which relies on a sequentially recursive adaptation process. As a consequence, it enables by nature the sequential access of HRTFs, e.g., for the virtualization of rotating sound sources. However, so far the direct access of a HRTF of an arbitrary azimuthal position, for a sudden change of the sound source position, is not possible without knowledge of neighboring HRTFs. To overcome this limitation, we present a new implementation strategy enabling a flexible, i.e., a sequential and direct, access of `Perfect Sequence LMS based HRTFs'. We will prove that we obtain numerically the same high-quality HRTFs as before, without increase of storage requirements, and a significantly reduced computational complexity.