Speech recognition using DNNs, script to create features, use kaldi for forced alignment and training networks for articulatory feature recognition