Content area
Full text
Abstract: The purpose of this paper is to present and compare the main techniques and methodological approaches aimed to improve audio discrimination in sound localization experiments. As reviewed from the literature, perceptual feedback training procedures provide listeners with the proper auditory/visual feedback concerning the correct sound source location. This type of feedback training can significantly improve localization accuracy, as well as the listener's ability to discern between the sound sources situated in the front and rear hemi-field, reducing the incidence of reversal errors (a very common situation in which the listener perceives a sound coming from the front as emerging from the back and vice versa).
For instance, in a training session, the listener indicates the perceived location of the sound source and receives immediate visual and audio feedback consisting in the presentation of the correct direction of the audio signal and instant playback of continuous streams of the corresponding sound source location. These results are long-term, suggesting that even short periods of training can enhance audio localization performance under 3D sound conditions, using non-individualized Head Related Transfer Functions (in the case of virtual auditory displays).
In this paper we focus on the examination and investigation of the role of perceptual feedback training under controlled conditions in both free-field (using loudspeakers as the main audio rendering channel) and in virtual auditory displays that employ headphone-presented sound. The results obtained from this study will build the basis of a methodology aimed to improve 3D sound localization in the case of the visually impaired people.
Keywords: Sound localization; 3D sound; virtual auditory display; training; perceptual learning.
I. INTRODUCTION
In order to decode the direction of a sound source in space, the human auditory system uses a set of binaural cues (for localization in the horizontal plane) and monaural cues (for vertical localization and front-back disambiguation), the latter being generally referred to as the Head Related Transfer Function (HRTF). As the HRTFs depend on the anatomical features of the listener (pinna, head and torso), they are significantly individualized for each person apart. The processing method of a 3D sound consists of synthesizing an audio signal with the corresponding pair of HRTFs (for the leftand for the right ear) of a given position in both the horizontal...