The intelligibility of whitened and amplitude compressed speech in a multitalker background.

Abstract

This experiment determined the effects of amplitude compression on speech intelligibility when both a target speech signal and a competing message were whitened and amplitude compressed. The target CNC discrimination words were electrically mixed with a competing message composed of five talkers. This composite signal was presented to normal hearing subjects in four ways: unmodified, whitened, whitened pplus 3:1 amplitude compression and whitened plus 10:1 amplitude compression. Discrimination functions were obtained for the CNC material by varying the signal-to-competition ratio. The unmodified and whitened speech yielded comparable discrimination functions, but reduced discrimination scores were obtained with the whitened lus compressed speech. However, the reduction in speech discrimination for the whitened plus compressed speech was slight and was most evident when the target signal and the competing background were at the same intensity.