Improved transition models for cepstral trajectories

Abstract

We improve on a piece-wise linear model of the trajectories of Mel Frequency Cepstral Coefficients, which are commonly used as features in Automatic Speech Recognition. For this purpose, we have created a very clean single-speaker corpus, which is ideal for the investigation of contextual effects on cepstral trajectories. We show that modelling improvements, such as continuity constraints on parameter values and more flexible transition models, systematically improve the robustness of our trajectory models. However, the parameter estimates re-main unexpectedly variable within triphone contexts, suggesting interesting challenges for further exploration.