Short-Time Spectrum Analysis of Speech