Detection and Recognition Thresholds of Environmental Sounds in Noise

Abstract

This study examines detection and recognition thresholds for environmental sounds in the presence of noise, with and without specific expectations about the targets. Human listeners were presented with a selection of everyday sounds masked by noise at initially increasing and later decreasing local signal-to-noise ratios. Participants had to indicate if they either detected or recognized the sound in the masking noise. The targets were selected from a database of environmental sounds. A previous perceptual similarity experiment on this database led to a separation in three classes that might be interpreted as predominantly tonal, pulsal, or noisy. We therefore separated the targets in these three groups. The resulting pattern of detection and recognition thresholds, with and without previous exposure to the target, suggest an 8 dB benefit for recognition and a 2dB benefit for detection. For repetitive sounds the detection benefit increased to 7 dB. The overall pattern of results provides support for the suggestion that sound recognition may be a combination of checking the presence of expected targets and a signal driven search in case of an unexpected target. The detection and recognition thresholds, suggest that human auditory perception might indeed employ different strategies for detecting tonal, pulsal, and noisy sounds.


Back to Friday Posters