Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition | IEEE Conference Publication | IEEE Xplore