Fast measurement of individual head-related transfer functions

Richter, Jan-Gerrit; Fels, Janina (Thesis advisor); Jax, Peter Johannes (Thesis advisor)

Berlin : Logos Verlag Berlin GmbH (2019)
Book, Dissertation / PhD Thesis

In: Aachener Beiträge zur Akustik 30
Page(s)/Article-Nr.: 1 Online-Ressource (III, 151 Seiten) : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2019

Abstract

While binaural technology applications gained in popularity in recent years, the majority of applications still use non-individual Head-Related Transfer Functions (HRTFs) from artificial heads. These datasets enable a reasonably good spatial localization which works especially well when using an additional visual cue. However, certain applications, for example research of spatial hearing or hearing attention, require an physically exact and realistic binaural signal. Moreover, it was shown in many experiments that there is a substantial gain from the use of individual HRTFs, for example in localization tasks. The limiting factor that prohibits the widespread use of individual HRTFs is the acquisition of such data. A substantial hardware requirement obstructs a more universal usage. Even for institutions that allow individual measurements, the measurement time that is required, and that the subjects are required to remain motionless made most measurements unfeasible in the past. This time requirement has recently been reduced by the use of parallelization in the measurement signal which lead to the development of fast measurement systems capable of acquiring individual and spatially dense HRTF. This thesis provides a objective and subjective evaluation of such a system that is designed with the goal of little disturbance of the measurements in mind. The construction is detailed, followed by both an objective and subjective evaluation. A detailed investigation into additional distortion of the sound field introduced by the system itself is presented and it is shown that the system performs comparably to a conventional system in terms of sound source localization. Furthermore, a method is introduced and evaluated to further reduce the measurement time by using continuous rotation during the measurement. This method is used to reduced the measurement duration from eight minutes to three minutes without audible differences. The introduced methods are also used to reducing additional errors from subject movement. It is shown that this movement can be reduced by a visual feedback system to a level that can be compensated efficiently.