Abstract:
In one of the earliest experiments concerning comodulation masking release (CMR), Hall et al. [J. Acoust. Soc. Am. 76, 50--56 (1984)] measured the detectability of a sinusoidal signal in the presence of a bandpass-noise masker centered at the signal frequency. Thresholds for noise maskers multiplied by a low-pass noise (``comodulated'' noise) were generally lower than for random noise maskers (``noncomodulated'' noise). In the literature it is generally assumed that the release is a consequence of an across-channel process. However, it is also argued in the literature [e.g., Schoonefeldt and Moore, J. Acoust. Soc. Am. 85, 273--281 (1989)], that some of the release might be due to a single-channel process, where threshold is determined by a change of the amplitude statistics within one critical band. Simulations with a modulation-filterbank model suggested by Dau et al. [J. Acoust. Soc. Am. 97, 3273(A) (1995)] were performed and compared to experimental data. The effects of varying masker bandwidth, center frequency, modulator bandwidth, and signal duration were examined. It is shown that a single-channel analysis accounts quantitatively for the CMR, indicating that across-channel processing is not necessary for simulating results from this kind of experiment.