Acoustic modeling with contextual additive structure for HMM-based speech recognition | IEEE Conference Publication | IEEE Xplore