Speech Recognition Engine using ConvNet for the development of a Voice Command Controller for Fixed Wing Unmanned Aerial Vehicle (UAV)

Current interfaces between humans and UAVs do little to recognize the human affinity for verbal communication or the accepted (and empirically effective) practice of guiding aircraft through verbal commands. Incorporating speech recognition into a technology solution is one way to attempt to determine what, exactly, was said by a human. Speech Recognition is an end to end process of mapping audio speech data to textual sentences or key phrases. The paper presents the implementation of the speech recognition engine for a voice command controller for fixed-wing Unmanned Aerial Vehicle (UAV) using a deep Convolutional Neural Network that process and classifies voice samples. The architecture was an adaptation of an image processing CNN, programmed in Python using Keras model-level library and Tensor Flow back-end. The original contribution of the paper is: the training of the CNN with a set of recordings in the English Language for basic Fixed-Wing UAV Maneuver commands like backward, down, follow, forward, go, left, off, on, right, stop, up, and addition of some commands for future purposes such as tree and visual. The proposed speech recognition engine is intended for the development of a voice command controller for Fixed Wing UAV.