Generative Kernels and Score-Spaces for Classication of Speech: Progress Report iii

May is is the third and nal progress report for Project /// (Generative Kernels and Score Spaces for Classiication of Speech) within the Global Uncertainties Programme. is project combines the current generative models developed in the speech community with discriminative classiiers. An important aspect of the approach is that the generative models are used to deene a score-space that can be used as features by the discriminative classiiers. is report discusses progress in three areas. First, eeciently computing segmental features for segments that begin or end at adjacent times. Second, nding the exact word error for discriminative training of acoustic models. ird, training innnite log-linear models with a criterion for the whole model. Experiments use a log-linear model with segmental features derived from a hidden Markov model with neural-network output distributions.