A computational auditory masking model based on signal-dependent compression. I, Model description and performance analysis

In the present study, a novel computational model is proposed, which can efficiently emulate many aspects of the monaural signal processing of the auditory system. The model employs the concept of Signal-Dependent Compression (SDC), according to which the auditory system is assumed to perform a compression of the input signal's dynamics, controlled by an operating-point signal, which is directly derived from the input signal itself. This model is characterized by a simple, analytically defined, and flexible general structure, which employs a small number of free parameters and allows for a computationally very efficient realization. The model seems to be useful in describing various aspects of psychoacoustical masking (simultaneous and forward masking) and shows significant physiological relevance (e.g., auditory adaptation, peripheral compression at critical frequency). Considering possible audio and speech processing applications, it is shown that the SDC approach can be regarded as an extension of widely used dynamic range control systems and in addition, produces a modulation transfer function similar to the modulation spectrum of speech. In an accompanying paper the model's performance in describing psychoacoustical masking data is investigated by comparing its output to results obtained from established psychoacoustical masking studies.