Multichannel audio upmixing based on non-negative tensor factorization representation | IEEE Conference Publication | IEEE Xplore