TEO-Pitch based classification of stressed speech under G-Force

Stressed speech recognition is a challenging work. One approach for addressing the degradation is to utilize front-end stress classification to direct a stress dependent recognition algorithm which separately models each speech production domain. Since G-Force has a direct physical impact on human speech production, G-Force induced stress is differ from stresses induced by psychological or perceptual factors which have been widely studied. So far, there is little research about classification of speech under G-Force. This study proposes a new approach based on several pitch features for the classification of G-Force/Normal speech. Stressed speech database under G-Force with two speakers has been collected in an aero-flight simulator, and the speaker dependent and multi-speaker average classification rate are 93.3% and 85.8% respectively.