1: \begin{abstract}

2: \Waspaa{Algorithms for} Blind Source Separation (BSS) of acoustic signals \Waspaa{require} efficient and fast converging optimization strategies to adapt to \Waspaa{nonstationary signal statistics and time-varying acoustic scenarios}. In this paper, we \Waspaa{derive} fast converging update rules \Waspaa{from a negentropy perspective, which are} based on the Majorize-Minimize (MM) principle and eigenvalue decomposition. The presented update rules are shown to \Waspaa{outperform competing state-of-the-art methods} in terms of convergence speed at a comparable runtime \Waspaa{due to the restriction to unitary demixing matrices. This is} demonstrated by experiments with recorded real-world data.

3: \end{abstract}