Effects of whitening and peak-clipping on speech intelligibility in the presence of a competing message.

The purpose of this experiment was to determine the effects of peak clipping on speech intelligibility when both a target speech and a competing message were simultaneously peak-clipped. A competing message composed of 5 talkers was electrically mixed with CNC discrimination words. This composite signal was presented to normally hearing subjects in three ways: unmodified, whitened, and whitened and peak-clipped. Discrimination functions were obtained for the CNC material by varying the signal-to-competition ratio. Under these conditions, essentially identical discrimination functions were yielded by the unmodified and whitened speech, whereas substantially reduced discrimination scores were obtained with the whitened/clipped speech. These results would suggest that speech intelligibility is reduced by whitening and peak clipping when more than one talker is present. This is true even though earlier studies have shown that whitening and peak clipping do not reduce speech intelligibility when only a single talker is present. Such a finding has implications for wearable amplification.