Musical pitch of interrupted white noise

Abstract

The pitch of periodically chopped white noise has been the object of several investigations in the literature. Because of the essentially flat spectrum of such a sound, any musical pitch effects are difficult to account for with a place model. Previous investigations, mostly involving pitch matching experiments, have left some doubt whether observed pitch effects were indeed factual or artifactual [Pollack, J. Acoust. Soc. Am. 45, 237 (1969)]. A series of musical interval identification experiments was performed using interrupted gaussian wide-band noise whose interruption rate was given by the notes to be presented. Intensities were 30 dB SPL, the noise on-time fraction was 0.5 and gating frequencies varied between 100 and 300 Hz. In this range, the pitch confusion data are well described by a simple Gaussian decision model whose variances are somewhat larger than the data scatter in previously reported pitch matching experiments. However, pitch identification performance with chopped noise is much worse than the same subjects' identification performance with pure tones or other periodic signals, which seems to indicate that, at least from a musical standpoint, we are dealing with a rather weak effect. [Work supported by the National Institutes of Health, grant 1RO1 NS 11680-01.]