In this paper the compact representation of the head-related transfer functions (HRTFs) or the Head-Related Impulse Responses (HRIRs) based on singular value decomposition (SVD) was investigated, focusing on the difference in the parameters to construct the average vectors and the covariance matrices in two points. One of them is on what the derived eigenvectors reflect the properties of HRTFs and/or HRIRs. As a result, high correlation between the parameters concerning SVD and the amplitude of HRTFs was found, and the high correlation was obtained in the wide region except the contralateral side. The second investigation is on the required number of the HRTFs or the HRIRs to construct the average vectors and the covariance matrices. It was found that the number of HRTFs are decreased to about 1/4 of the whole directions for the used HRTFs. Among three conditions of the HRIRs, the amplitude of HRTFs, and the log-amplitude of HRTFs, it was also shown that the amplitude of HRTFs is the most effective to construct the parameters of the SVD.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.