Robust speech pulse detection using adaptive noise modelling andnon-stationarity measure

摘要

The problem of speech pulse detection with additive noise at SNRas low as 0 and -6 dB is addressed. The noise is assumed to bereasonably stationary and correlated. Three techniques have beenexamined: firstly, the autoregressive analysis of noise; secondly,spectral density comparison; and finally nonstationarity measure