Model A: Auditory segregation model

  

Figure: Model A: an auditory segregation model.

The auditory segregation model, shown in Fig. , consists of three parts: (a) an auditory filterbank, (b) separation block, and (c) grouping block. The auditory filterbank is constructed using a gammatone filter as an ``analyzing wavelet.'' The separation block uses physical constraints related to heuristic regularities (ii) and (iv) proposed by Bregman [6]. The grouping block synthesizes each separated parameter and then reconstructs the extracted signal using the inverse wavelet transform.