Learning to Detect and Identify Acoustic Environments From Reflected Sound

Abstract

Reflected sounds are often treated as an acoustic problem because they produce false localization cues and decrease speech intelligibility. However, their properties are shaped by the acoustic properties of the environment and therefore are a potential source of information about that environment. The objective of this study was to determine whether information carried by reflected sounds can be used by listeners to enhance their awareness of their auditory environment. Twelve listeners participated in two auditory training tasks in which they learned to identify three environments based on a limited subset of sounds and then were tested to determine whether they could transfer that learning to new, unfamiliar sounds. Results showed that significant learning occurred despite the task difficulty. An analysis of stimulus attributes suggests that it is easiest to learn to identify reflected sound when it occurs in sounds with longer decay times and broadly distributed dominant spectral components.

Notes

Portions of this work were presented at the Psychonomics Conference in Houston, Texas, in November 2006.

^aDecay time is defined as the time required for the intensity of a sound to decrease 20 dB from its peak value.

¹It should be cautioned that equating the scale of a wavelet to a particular frequency is slightly inaccurate, because a wavelet is a finite complex waveform and consists of multiple frequencies.