Paper Title
Dynamic Modeling of Speech by Self-Organizing Maps
Abstract
This paper tended to the issue of modeling the dynamics of speech based on temporal self-organizing neural
networks (SOMs).Starting from the premise that the dynamic conduct of the phonetic constituents can be recognized in
characteristic discourse at the neural level, this work examined the likelihood of separating the highlights of
phones/phonemes utilizing dynamic SOMs. A version of temporal SOM was suggested that demonstrated the capability to
incorporate the phones/phonemes dynamic features and furthermore indicate the component trajectories as they show up in
the feature space. The simulation results proved the potential offered by this version of temporal SOMs to model the
dynamic features and trajectories for both the individual phones/phonemes and words. The present approach isconsistent
with recent findings in the field of cognitive neurodynamics, and can be extended as well for modeling brain dynamics in
other linguistic studies.
Index terms - Speech Modeling, Dynamic Modeling, Neural networks, Self-Organizing Maps, Semantic Modeling, Time
Series Modeling.