Auditory Spatial Saliency and Its Effects on Perceptual Noisiness

Environmental noise affects significantly our health and quality of life. Simple techniques, such as A-weighted decibels, have been applied commonly to the real issues of environmental noise. More elaborate techniques, considering mechanisms in the central nervous system, have also been developed continuously and approved by the international organization of standardization (e.g., ISO 532-3; hereafter, ISO loudness model). These techniques have advanced our knowledge of perceptual noisiness, but still have some limitations to account for a variety of psychophysical phenomena and our empirical experiences in acoustic engineering. Here, we propose that perceptual noisiness can be explained better by considering auditory attention. Attention driven by sensory input has been modeled originally as “saliency” in vision. This algorithm has also been applied to capture spectral-temporal dynamics of auditory attention (hereafter, spectral saliency). It has been suggested that the central auditory system contains two pathways identifying what and where a sound source is. The above spectral saliency corresponds only to the what-pathway. We therefore created a new auditory spatial saliency model to capture attentional effects along the where-pathway based on an algorithm of horizontal sound localization. We found that our spatial saliency model accounted for perceptual phenomena that cannot be explained by the ISO loudness model. Furthermore, the prediction of perceptual noisiness of environmental sounds (driving sounds of passenger cars) was improved significantly by integrating spatial saliency with ISO loudness. We conclude that spatial saliency can be used to capture sound features affecting perceptual noisiness in everyday life.