ASA 127th Meeting M.I.T. 1994 June 6-10

4pPP26. Detection and recognition of familiar sounds.

Kim S. Abouchacra

Pamela A. Mundis

Ellen C. Haas

U.S. Army Res. Lab., Human Res. and Eng. Directorate, Aberdeen Proving Ground, MD 21005-5425

Tomasz R. Letowski

Laurie Myers

Penn State Univ., University Park, PA 16802

An exploratory study was conducted to determine which acoustical components influence the detection and recognition of 30 familiar sounds (FS). The sounds were categorized as human-, animal-, environment-, or object-producing sounds. Initially, detection thresholds for 20 normal hearing subjects were measured for the 30 FS. Following detection measures, subjects were trained to recognize FS that were presented at a comfortable listening level (40 dB HL). Recognition thresholds were then gathered, beginning at each sound's detection threshold. Specifically, at detection threshold, a sound was presented and subjects were required to identify the sound. If they could not recognize the sound, listeners were asked to describe ``what'' was heard (i.e., temporal or spectral components). In increments of 2 dB, the sound was presented and subject's response recorded. This procedure continued until the level of the sound reached 40 dB HL (training level). Results revealed that the decibel range between detection and recognition averaged 6--10 dB, depending on the sound. At detection threshold, no sounds were recognized; however, subjects predominately heard temporal components of the sound. As levels increased above detection, subjects heard spectral components of the sound prior to its correct recognition. Results suggest that temporal information may play a major role in the detection of sounds, whereas a combination of temporal and spectral information is required for the recognition of familiar, everyday sounds.