C. Formby
Dept. of Otolaryngol.---HNS, Johns Hopkins Univ., P.O. Box 1737, Baltimore, MD 21203
M. G. Heinz C. E. Luna
Johns Hopkins Univ., Baltimore, MD 21218
A new detection model can predict masked detection thresholds for a range of signal bandwidths (W=62--6000 Hz) and durations (T=10 to 500 ms) for noise signals centered at 2500 Hz in an uncorrelated 500-ms, 6000-Hz-wide masker (N[sub 0]=53 dB SPL/Hz). The new model effectively takes into account relative cues resulting from temporal and spectral differences between the signal and masker to supplement energy detection. These relative cues are (1) greatest when the durations and spectra of the signal and masker are most disparate and (2) eliminated when the signal and masker are matched temporally and spectrally. The relative timing and spectral shape cues combine independently and equally to enhance detection by as much as 10--12 dB relative to energy detection alone. The derivation of the new detection model will be explained and predictions from the new model will be compared with experimentally measured masked detection thresholds. The overall agreement between predicted and measured thresholds is excellent (r=0.95). [Research supported by NIH.]