jupiter-gloveVoice is one of the most important forms of communication for humans. It transmits a lot of information and can be used while our hands are busy performing other delicate tasks as the ones that are performed in space, for instance in a Extra-Vehicular Activity. That way, we want to focus on developing an efficient voice commands interface.
This project is solving the SpaceGloVe: Spacecraft Gesture and Voice Commanding challenge. Description
This projects focuses on creating an efficient voice command interface but may also include basic motions commands.
To achieve an efficient voice command interface we want to divide the interface in two parts:
The voice capture module, this can be in the user's arm and can use an optional headset for a better capture of the voice commands, this module will record the voice and extract features from it. This characteristics will be sent to the processing module. This module will also capture the gesture commands. As the characteristics, and not the full voice signal is the one being sent, this will speed up and make more efficient the communication with the processing module. In the prototype developed this weekend this module will consist of a mobile phone or tablet running the Android system. Most code for this module will be written in Java, but some may be written in C if it is found to be more efficient.
The processing module will take the characteristics extracted from the voice signal by the voice capture and process with an automatic speech recognition system to turn them to text. This text will be used to issue the command. To implement this module this weekend we are planning to use the CMU Sphinx Toolkit.
License: BSD 3-Clause "New" or "Revised" License (BSD-3-Clause)