Abstract:

Psychoacoustic experiments are conducted to investigate the broadening and the evenness of auditory source width （ASW） when uncorrelated noise signals are reproduced by a pair of virtual loudspeakers respectively located at ±45°, ± 60°, ± 75° and ±90°, and by a pair of virtual loudspeakers and a real central loudspeaker simultaneously. Results show that： when signals are reproduced only by a pair of virtual loudspeakers, ASW can be broadened to a maximum value of about ± 70° for octave noises with a central frequency varying from 250 Hz to 1 kHz, but the maximum reaches about ±50° for pink noises. When the angle between two virtual loudspeakers is wide, ASW can not be evenly broadened and it becomes empty in middle and strong in sides. When a pair of virtual loudspeakers and a real central loudspeaker replay simultaneously, the sound pressure level of the central loudspeaker relative to virtual loudspeakers greatly affects the ASW and its evenness. ASW can be evenly broadened to a wide range by adjusting the sound pressure level of the central loudspeaker. Furthermore, theoretical calculation and test results show that there is a negative linearity between ASW and IACC （the maximum of inter-aural crosscorrelation function） only when IACC0 （the inter-aural cross-correlation value when r = 0） is greater than 0, whereas for all kinds of replay signals, the negative linearity always exists between ASW and IACC0.

Key words: auditory source width, virtual auditory display, inter-aural cross-correlation