Implementation of improved 3D interpolation technique on HRIR models using PCA
Hugeng Hugeng, Irfan Naufal · IOP Conference Series Materials Science and Engineering · 2019
Implementation of head-related impulse responses (HRIRs) in the present can be felt and experienced by most people. Three-dimension sound using HRIRs has been applied in various fields, such as voice virtualization, music, film, and also for video games. However, the amount of data available in the HRIR database required for 3D sound is substantial, hence making HRIR application more efficient even in computational systems that are limited in memory, requires a dimension reduction method for the HRIR data. This research used Principal Component Analysis (PCA) modelling method using HRTF PKU-IOA database. We analysed three interpolation techniques on the HRIR models; those are bilinear rectangular, bilinear triangular and tetrahedral interpolation technique. The best result of interpolation technique is by applying tetrahedral interpolation on PCA modelling of HRIR data, that has mean percentage error of 7.5034% from minimum phase HRIRs and spectral distortion of 4.7677 dB from magnitude HRTFs.