This study presents an auditory processing model that accounts for the perceptual phenomenon of comodulation masking release (CMR). The model includes an equalization-cancellation (EC) stage for the processing of activity across the audio-frequency axis. The EC process across frequency takes place at the output of a modulation filterbank assumed for each audio-frequency channel. The model was evaluated in three experimental conditions: (i) CMR with four widely spaced flanking bands in order to study pure across-channel processing, (ii) CMR with one flanking band varying in frequency in order to study the transition between conditions dominated by within-channel processing and those dominated by across-channel processing, and (iii) CMR obtained in the “classical” band-widening paradigm in order to study the role of across-channel processing in a condition which always includes within-channel processing. The simulations support the hypothesis that within-channel contributions to CMR can be as large as 15dB. The across-channel process is robust but small (about 2–4dB) and only observable at small masker bandwidths. Overall, the proposed model might provide an interesting framework for the analysis of fluctuating sounds in the auditory system.

This study presents an auditory processing model that accounts for the perceptual phenomenon of comodulation masking release ͑CMR͒. The model includes an equalization-cancellation ͑EC͒ stage for the processing of activity across the audio-frequency axis. The EC process across frequency takes place at the output of a modulation filterbank assumed for each audio-frequency channel. The model was evaluated in three experimental conditions: ͑i͒ CMR with four widely spaced flanking bands in order to study pure across-channel processing, ͑ii͒ CMR with one flanking band varying in frequency in order to study the transition between conditions dominated by within-channel processing and those dominated by across-channel processing, and ͑iii͒ CMR obtained in the "classical" band-widening paradigm in order to study the role of across-channel processing in a condition which always includes within-channel processing. The simulations support the hypothesis that within-channel contributions to CMR can be as large as 15 dB. The across-channel process is robust but small ͑about 2 -4 dB͒ and only observable at small masker bandwidths. Overall, the proposed model might provide an interesting framework for the analysis of fluctuating sounds in the auditory system.

No abstract

The relative contributions of within-channel and across-channel processes to perceptual comodulation masking release (CMR) were investigated in the framework of an auditory processing model. A generalized version of the computational auditory signal processing and perception model [CASP; Jepsen et al., J. Acoust. Soc. Am. 124, 422-438 (2008)] was used and extended by an across-channel modulation processing stage according to Piechowiak et al. [J. Acoust. Soc. Am. 121, 2111-2126 (2007)]. Five experimental paradigms were considered: CMR with a broadband noise masker as a function of the masker spectrum level; CMR with four widely spaced flanking bands (FBs) varying in overall level; CMR with one FB varying in frequency and level relative to the on-frequency band (OFB); CMR with one FB varying in frequency; and CMR as a function of the number of FBs. The predictions suggest that at least three different mechanisms contribute to overall CMR in the considered conditions: (1) a within-channel process based on changes in the envelope characteristic due to the addition of the signal to the masker; (2) a within-channel process based on nonlinear peripheral processing of the OFB's envelope caused by the FB(s); and (3) an across-channel process that is robust across presentation levels but relatively small (2-5 dB).