Unsupervised speaker adaptation based on hierarchical spectral clustering | IEEE Journals & Magazine | IEEE Xplore