Application of Discriminant Analysis Techniques to G-Stress Analysis in Computer Voice Decoding

摘要

The Statistical Package for the Social Sciences (SPSS) Discriminant Analysis Routines were applied to speech data obtained at various gravity levels. The data were created on the Air Force Medical Research Lab's Centrifuge at Wright-Patterson Air Force Base. They were then digitally sampled and Fourier transformed to 64 frequency bands which were converted to two separate files; one the 16 frequencies used by Guyote and Sisson; one the 19 frequencies used by Threshold Technology, Inc. These files were then further processed: summing the energy across time slices, summing across frequency bands, and averaging across frequency bands. Statistical results showed that G-Stress seemed to be moving energy across time slices, more than shifting across frequencies. (Author)