Phoneme contrasts with two, same-talker speakers

Abstract

People often have difficulty hearing speech in the presence of concurrent conversations. This well-known cocktail-party effect can be parsed into energetic and informational masking effects. The purpose of this study was to measure and model the effects of energetic and informational masking that occur when two words, spoken by the same talker are heard at the same time. The word identification test used in the experiments was the Modified Rhyme Test (MRT). The MRT was used as both the stimulus and the masker, which afforded a multitude of consonant contrasts. The phrases were presented monaurally at 75 dB SPL over Sennheiser HD-520 headphones to four normal hearing listeners. The independent variables included 30 pairs of MRT word lists spoken by three male and three female talkers. The dependent variable was the percent correct identifications of the two consonants. Listeners performed at 94% correct for the first word choice and 72% correct for the second word choice. The distribution of errors was analyzed by place of articulation, manner of articulation, and speech-to-speech-ratio of the phoneme pairs and compared to articulation index predictions for speech intelligibility.