Blind Separation of Instantaneous Mixtures of Non Stationary Sources

Most source separation algorithms are based on a model of stationary sources. However, it is a simple matter to take advantage of possible non-stationarities of the sources to achieve separation. This paper develops novel approaches in this direction, based on the principles of maximum likelihood and minimum mutual information. These principles are exploited by e#cient algorithms in both the o#-line case (via a new joint diagonalization procedure) and in the on-line case (via a Newton-like procedure). Some experiments are presented showing the good performance of our algorithms and evidencing an interesting feature of our methods: their ability to achieve a kind of super-e#ciency. The paper concludes with a discussion contrasting separating methods for non-Gaussian and non-stationary models and emphasizing that, as a matter of fact, `what makes the algorithms work' is ---strictly speaking--- not the non stationarity itself but rather the property that each realization of the source signals has a time-varying envelope