A new approach to time-frequency signal decomposition

A recently developed method for separating time-frequency disjoint signal components is based on the Wigner distribution (WD) and uses a time-frequency masking procedure followed by a signal synthesis step. A problem with this method is that the positioning of the time-frequency masks is done manually and can be difficult due to WD interference terms. Also, masking is ineffective if a WD interference term overlaps with the WD signal term to be isolated. The authors present an approach to time-frequency signal decomposition which uses an extended version of signal synthesis and does not require a masking procedure.<
>