Multilingual voice recognition using deep learning for human-drone interaction

Abstract

Over the past several years, human-drone interaction have got an important part from the scientific researchs. When interacting with a drone, humans take on many responsibilities. Their role is determined by the drone’s application and the amount of autonomy. The purpose of this work is to control the movement of the unmanned aerial vehicle (UAV) using a multilingual speech recognition system. For that, a deep neural network (DNN) is trained to recognize the user’s speech from many languages and then generate the desired control command. We conducted experiments involving participants giving voice commands in order to compare the effectivencess of each database. Hardware implementation of the designed system proof its high accuracy recognition and control simplicity.