Evaluation of Simultaneous Speech Detection Based on MFCC-DTW with Two-Stage Normalization

Abstract

In Air Traffic Control a serious safety risk is represented by undetected simultaneous transmissions from different airplanes. In this paper, we approach this issue through a speech analysis algorithm, which combines traditional Mel Frequency Cepstral Coefficients extraction, a new two-stage normalization and widely used Dynamic Time Warping. In this way, we were able to extend the simultaneous speech detection capability in Voice Communication Systems of Air Traffic Control. The results prove that this implementation is suitable for practical applications.