Towards self-personalizing headphones: Individual multi-driver-multi-receiver headphone impulse responses and...

Acquisition of individual head-related transfer functions (HRTFs) still requires specialized equipment, and simplification of that process remains a challenging research objective. A method for automated HRTF individualization relying on a self-personalizing headphone (Selfone) would represent an elegant solution. This article describes a Selfone prototype consisting of 20 drivers and 24 receivers. Further, the article describes an acoustic database consisting of datasets of Selfone impulse responses (SIRs) measured between the Selfone drivers and receivers and in-ear impulse responses (IRs) measured at the ear-canal entrance, from 100 participants and for seven Selfone wearing positions. The datasets are complemented by HRTFs measured from the same individuals. As expected, substantial differences between the in-ear IRs and HRTFs for approximately matched directions were found, indicating that a simple playback via individual Selfone drivers would create strong deviations from individual HRTF cues. This emphasizes the need for algorithms estimating driving functions required to maintain individual HRTF cues. Further, individual pinna morphology led to large variations in SIRs, indicating that at least some of the individual features have been captured by the SIRs. Being publicly available, the Selfone database aims to support the development of future algorithms that facilitate to estimate individual monaural HRTF cues from SIRs.