Modeling phones coarticulation effects in a neural network based speech recognition system

In this paper we have designed and implemented speech recognition models in phone recognition level to model phones coarticulation effects. We have inspired these models from two human cognitive systems: neocortex and hippocampus. In the model inspired from the neocortex the first step is a primary and coarse classification of inputs, then model adapts itself to contexts extracted from these primary recognitions and we classify inputs again according to their extracted context. In the model inspired form the hippocampus, previous contexts of inputs are used for better recognition, and in this way we use effects of previous phones of each input for better classification. Then we have designed and implemented a model with a structure of combination of two preceding models. Our models implementation showed 3.77% increase in accuracy of Persian phone recognition compared to a simple model that does not consider coarticulation effects.