Comodulation masking release as a function of type of signal, gated or continuous masking, monaural or dichotic presentation of flanking bands, and center frequency.

Thresholds were measured for detecting a signal centered in a narrow-band noise (NBN) masker (on-frequency band, OFB), for the OFB alone, and with two flanking bands (FBs) added to the OFB, one centered above and one below the OFB. The FBs were either correlated with the OFB or were independent and were presented either to the same ear as the signal plus OFB (monaural condition) or to the opposite ear (dichotic condition). The OFB and FBs were either gated with the signal, or were presented continuously. Three signal types were used: a pure tone; an NBN uncorrelated with the OFB; and an NBN correlated with the OFB. The signal was centered at 0.5, 2, or 6 kHz. Comodulation masking release was estimated either as the difference between threshold with the OFB alone and with the OFB plus correlated FBs [CMR(R-C)], or as the difference between thresholds using correlated and uncorrelated FBs [CMR(U-C)]. Although there were marked individual differences, positive CMR(R-C) values were found in all conditions for all three signal types. CMR(U-C) values were often larger than those for CMR(R-C), reflecting the fact that the uncorrelated FBs tended to produce interference effects, especially for the gated maskers, and at 6 kHz. Values of CMR were larger and more consistent across subjects for continuous than for gated maskers. For continuous maskers, the values of CMR tended to be smallest for the correlated-NBN signal. Results are discussed in terms of available cues and in terms of perceptual grouping mechanisms.