Unsupervised hierarchical structure induction for deeper semantic analysis of audio | IEEE Conference Publication | IEEE Xplore