Cocktail Party Effect

Abstract

Acousticians have long recognized that the cocktail party is a stimulating environment for inquiry into the basic problem of acoustics. The cocktail party provides a critical test of listening to a talker against extraneous sound sources. The “cocktail party effect” hence has been identified with situations which require selective directional listening in a multitalker environment. After appropriate operational research of cocktail parties, we brought the problem to the laboratory. Unfortunately, we were forced to reduce the problem to more sober proportions. We tested the intelligibility of recorded monosyllabic PB words against a babble of 1, 2, 4, or 7 independent talkers reading newspaper text. Two test conditions were examined: (1) test words were introduced stereophonically with one set of background talkers presented over one earphone and another set of talkers presented over the other earphone; and (2) test words were introduced with a single set of background talkers presented over a single earphone. The cocktail party effect was identified with the difference between the speech-to-background ratios required for 50% intelligibility under the two conditions. The effect ranged from 12 db with one back-ground voice to 6 db with seven background voices (in each ear).