摘要

In this paper, an improved sound image externalization approach is proposed by considering the essential differences between headphone listening and real-life listening to deal with the problem of in-head localization of sound images. Through simulating the virtual sound source in a given enclosure, the proposed externalization approach renders sound images out-of-head by exploiting a Binaural room impulse response (BRIR) simulation approach, which consists of a new hybrid approach for the acoustic Room impulse response (RIR) and the Head-related impulse responses (HRIRs). The hybrid RIR simulation approach comprises an improved image-source model for the early reflections by integrating the air absorption, and a Feedback delay network (FDN) for the late reflections. Moreover, the HRIRs are further integrated into each sound image in the early reflections to imitate the effect of listener. Psychoacoustic test results demonstrated that the proposed externalization algorithm offered significant out-of-head perception and more natural quality of sound images.

全文