Grouping constraints

The aim of the grouping constraints is to extract the desired signal from the noise-added signal using regularities (i) and (iii) proposed by Bregman. Therefore, the grouping block takes a solution for the problem of segregating two acoustic sources and applies to X_k(t), in which two acoustic signals exist in the same time region. In other words, it applies the solution to X_k(t), if either of the two physical constraints as shown below are satisfied[5].

1.

Harmonicity

When, the channel number k corresponds to an integer multiple of the fundamental frequency F₀, $n\cdot F_0,n=1,2,\cdots,N_{F_0}$. This constraint means that ``when a body vibrates with a repetitive period, its vibrations give rise to an acoustic pattern in which the frequency components are multiples of a common fundamental.''

2.

Common onset and offset

When onset and offset of X_k(t) for f₁(t) match those of X_k(t), corresponding to the fundamental frequency F₀. This constraint means that ``unrelated sounds seldom start or stop at exactly the same time.''