Inter-Talker Differences in Intelligibility for Two Types of Degraded Speech

Abstract. Are the acoustic-phonetic factors that promote highly intelligible speech invariant across different listener populations and listening environments? Researchers have taken two approaches to investigate differences in intelligibility for a variety of listener populations: examining how speaking style affects intelligibility, and examining how inter-talker differences influence intelligibility. Following the latter approach, we compared the intelligibility of talkers under cochlear implant (CI) simulation (n=200), and in speech mixed with babble (n=200) with their intelligibility under quiet listening conditions (n=200, reported by Karl & Pisoni, 1994). The stimuli consisted of 20 native English talkers producing 100 sentences which were processed to simulate listening with an 8-channel CI or mixed with multi-talker babble. For each condition, stimuli were presented to listeners in a sentence transcription task. The results indicated that the most intelligible talkers in quiet were not the most intelligible talkers under CI-simulation or in babble. Furthermore, listeners demonstrated a greater degree of perceptual learning with the CI-simulated speech compared with the speech mixed with babble. While some of the acoustic-phonetic properties were correlated with intelligibility in all conditions