A joint time-frequency model of auditory masking

Abstract

The perceptual phenomena of temporal and spectral masking (critical band masking) [B. C. J. Moore, Frequency Selectivity in Hearing (Academic, New York, 1986)] are related to the ear’s limited temporal and spectral resolution. This resolution is known from psychoacoustic experiments but is not typically incorporated in the time-frequency representations used in audio signal processing. A joint time-frequency distribution, modeling both critical band masking and nonsimultaneous (forward and backward) masking, is described here. Applications for such a distribution include audio compression and the modeling of room acoustics. The distribution is developed within the framework of Cohen’s class of generalized time-frequency distributions [L. Cohen, ‘‘Time-frequency distributions—A review,’’ Proc. IEEE 77, 941–981 (1989)] and is a development from the constant Q modal distribution [W. J. Pielemier and G. H. Wakefield, ‘‘A high resolution time-frequency representation for musical instrument analysis,’’ J. Acoust. Soc. Am. 98, 2382–2396 (1996)]. The design and implementation of this auditory model is detailed.