Error analysis of HRTFs measured with complementary (Golay) codes
Corey I. Cheng, Gregory H. Wakefield · The Journal of the Acoustical Society of America · 2001
Head-related transfer functions (HRTFs) characterize the spatial filtering properties of the external ear, and can be measured using various system identification techniques. These techniques generally assume that the system under study is time invariant (e.g., motionless). In a recent study, Zahorik presented empirical results concerning the effects of head motion on the accuracy of a popular system identification procedure based on pairs of binary stimulus signals called complementary, or Golay, codes [P. Zahorik, J. Acoust. Soc. Am. 107, 1793–1796 (2000)]. In this paper, we present a simple model for Zahorik’s work which handles not only the effects of head motion, but also measurement noise and dc bias as well. This theoretical formulation confirms the observations made by Zahorik to the extent that head motion can introduce significant errors in both the contour and fine structure of the measured HRTF log magnitude response. Further analysis of this model suggests why another popular system identification procedure based on maximal length sequences (MLSs) does not suffer from these errors. Finally, an unexpected finding is that these errors depend on the spectral shape of the specific Golay pair used during measurement. [Work supported by a grant from ONR.]