Intelligibility of selected passages from the Speech Intelligibility Rating (SIR) test

Two experiments were conducted to examine the intelligibility of 12 of the 72 passages of connected discourse prepared by Cox and McDaniel (1984, 1989) in the development of the Speech Intelligibility Rating (SIR) test. Intelligibility was assessed with a method-of-adjustment procedure in the presence of two maskers. One was a multi-talker babble with a variable S/N ratio environment that yields intelligibility scores that are potentially level-dependent because of the almost inevitable difference in speech intensity from passage to passage. The second was a signal-correlated noise with a constant S/N ratio environment that provides scores that are essentially level-independent. Two homogeneous subsets of nine passages each were identified that yield equivalent intelligibility scores. The outcome underscores the value of incorporating a signal-correlated noise masker that yields scores that are relatively unaffected by small differences in signal level among passages.