Sparse representation of phonetic features for voice conversion with and without parallel data | IEEE Conference Publication | IEEE Xplore