Independent component analysis for extraction of critical features from tongue movement ear pressure signals

The goal of this work is to discover and extract critical features from pressure signals in the aural canal resulting from actions (tongue movements) within the oral cavity. Its scope encompasses the identification of critical features of pressure signals sensed in the ear resulting from tongue motion and the development of algorithms and methodologies to extract features from sets of these signals. We report successfully isolating 15 components associated with four different tongue motions and clustering them into 3 groups based on similarity in characteristics. The components are consistently extracted with every repetition, irrespective of the start of the tongue movement, thus providing a venue of correlating the signal to the action without dependency on complicated algorithms identifying the start and endpoints of the signal. To our knowledge, this work is the first ever analysis of components of pressure signals in the ear canal associated with tongue movement. In future work, these findings are expected to lead to an entirely new generation of unobtrusive human-machine interface mechanisms.