Detection of stop landmarks using Gaussian mixture modeling of speech spectrum

Perception of speech under adverse listening conditions may be improved by processing it to incorporate properties of clear speech. It needs automated detection of stop landmarks and enhancement of bursts and transition segments. A technique for accurate detection of stop landmarks in continuous speech based on parameters derived from Gaussian mixture modeling of log magnitude spectrum, a voicing onset-offset detector, and a spectral flatness measure is presented. Applying the technique on sentences from the TIMIT database resulted in burst detection rates of 98, 97, 95, 90, and 73% at temporal accuracies of 30, 20, 15, 10, and 5 ms respectively.