(Deep) Neural Networks

This work continues in development of the recently proposed Bottle-Neck features for ASR. A five-layers MLP used in bottleneck feature extraction allows to obtai arbitrary feature size without dimensionality reduction by transforms, independently on the MLP training targets. The MLP topology – number and sizes of layers, suitable training targets, the impact of output feature transforms, the need of delta features, and the dimensionality of the final feature vector are studied with respect to the best ASR result. Optimized features are mpl yed in th ee LVCSR tasks: Arabic broadcast news, English conversational telephone speech and English meetings. Improv ment over s andard cepstral fe tures and probabilistic MLP features are shown for different tasks and differ nt neural net input representations. A significant improvement is observed when phoneme MLP training targets are replaced by phoneme states and when delta features a e added. Index erms— Bottle neck, MLP struc ure, fea ures, LVCSR