Head Related Transfer Function Interpretation Method Based On Spatial Perception

The virtual reality(VR)technology has brought us an immersive experience,it's going to be one of the key technologies of the multimedia area,and lead the development of it.Most of the VR technologies used stereo to reconstruct the three demonstration(3D)sound field,the reconstructed signals are limited in a small area,which cannot completely match the necessary of the 3D video content.As one of the internationally recognized 3D audio generation methods,the head related transfer function(HRTF)technology has been widely used,but the interpolation error always affect the accuracy of the signals' spatial position.This paper conducted the research about the spatial perception mechanism at first,and then conducted the research on HRTF interpretation method.The works are as follows.Research on spatial perception mechanism.Most of the existed measurements and analysis about the perception mechanism are limited in a small area or planes,and most of them are about the azimuth minimum audible angle(AMAA),it is unable to reflect the completely perception characteristics of the human ear.This paper designed a subjective listening test based on psychoacoustics,built a testing signal database through collecting signals at difference locations in an anechoic room,conducted the elevation minimum audible angle(EMAA)tests with the headphones.The azimuth-elevation perception database were made through a joint analysis on the measured EMAA data and known AMAA data,the database could be used in virtual sound source processing and 3D audio generating.Research on HRTF segmented interpretation method based on spatial perception.Most of the existed HRTF interpretation methods ignore the connection between HRTF data and its spatial location,the structure characteristics of the HRTF data itself is not being valued too,the error between interpretation and original HRTF caused the inaccuracy of the signals' spatial position.This paper is based on the spatial perception characteristic,a judgement based on perception characteristic would be made before the HRTF interpretation,if the interpreting HRTF is within the perception area of the existed HRTF data,the existed HRTF data would be considered as the interpretation HRTF data directly,if not,the interpretation HRTF data would be made by segmented interpretation method.The segmented interpretation method used back propagation neutral network to generate the two ears' delay threshold,used bilinear interpretation method to generate the valid HRTF data,and used original HRTF data to generate the noise signal,the final interpretation HRTF data were stitched after these three calculations.This method does not only take the perception characteristic into consideration,but also designed a segmented interpretation method for the HRTF data,which aims to reduce the error during the interpretation as much as possible.Several experiments were carried out in this paper,the results show that,as to the bilinear interpretation method,the method that presented in this paper has the lower error and the better spatial positioning accuracy.