Dependency of the Finite-Impulse-Response-Based Head-Related Impulse Response Model on Filter Order

摘要

Various approaches have been reported on HRIR modeling to lighten the high computation cost of the 3-D audio systems without sacrificing the quality of the rendered sounds. The performance of these HRIR models have been widely evaluated usually in terms of the objective estimation errors between the original measured HRIRs and the modeled HRIRs. However, it is still unclear how much these objective evaluation results match the psychoacoustic evaluations. In this research, an efficient finite-impulse-response (FIR) model is studied as a case study which is essentially based on the concept of the minimum-phase modeling technique. The accuracy dependency of this modeling approach on the order of FIR filter is examined with the objective estimation errors and the psychoacoustic tests. In the psychoacoustic tests, the MIT HRIR database are exploited and evaluated in terms of sound source localization difference and sound quality difference by comparing the synthesized stimuli with the measured HRIRs and those with the FIR models of different orders. Results indicated that the measured hundred-sample-length HRIRs can be sufficiently modeled by the low-order FIR model from the perceptual point of view, and provided the relationship between perceptual sound localization/quality difference and the objective estimation results that should be useful for evaluating the other HRIR modeling approaches.