This document presents a new method called multi-view spectral clustering (mSC) for discovering multiple non-redundant clusterings from multi-view data. mSC optimizes both a spectral clustering objective to measure cluster quality and a Hilbert-Schmidt independence criterion regularization to measure redundancy between clusterings. It can discover clusters with flexible shapes while simultaneously learning the subspace for each clustering view. The method is evaluated on several datasets and is shown to outperform other multi-view clustering algorithms in terms of normalized mutual information.