This work aims at deriving a minimum required resolution for optimization of head-related transfer functions (HRTFs). It builds on existing metrics, used to numerically evaluate HRTF differences, as well as on a model estimating just noticeable differences (JNDs) for uni-lateral variation of HRTFs. Integrating this model, as well as descriptors for both monaural and binaural cue differences, a three-alternative forced choice experiment is set up to investigate JNDs for bi-lateral variation of HRTF sets. Rather than introducing manual changes to the spectra, an exchange between magnitude spectra of generic HRTF sets is employed, while controlling for multiple conditions related to the descriptors. The probability of distinguishing between the stimulus pairs is linearly modeled using different subsets of numerical descriptors. A model integrating two monaural descriptors, ‘issd’ and ‘mfcd’, achieves the best performance, compared to the rest. It shows a tendency for slight improvement when combined with an estimate of the detectability of changes in interaural cross-correlation.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22061
(496KB)
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 Join the AES. If you need to check your member status, login to the Member Portal.
