Effects of Coarticulation, Prosody, and Noise Freshness on the Intelligibility of Digit Triplets in Noise

Abstract

In a number of European countries, a functional self-test to screen for hearing impairment is available via telephone and the Internet. The tests estimate speech-reception thresholds using an adaptive procedure in which digit triplets are presented at varying signal-to-noise ratios. In different languages, the stimuli were created either with or without coarticulation; and some implementations use fresh noise samples, while others do not. The present investigation concerns the influence of coarticulation, prosody, and noise freshness on measured thresholds. We performed a laboratory study using 12 normal-hearing listeners. In a blocked design we compared speech-reception thresholds for conditions with and without fresh noise tokens. In each block we used three types of triplets: with coarticulation and prosody, with neither, and without coarticulation but with prosody. Thirty-six thresholds were recorded per subject, and they were analyzed using analyses of variance. The results showed no significant differences among the three triplet conditions. The freshness of the noise did not affect thresholds when, at least, a fresh noise token was used per threshold estimate (23 presentations). Scores dropped significantly when a whole experimental block was performed with a single noise token.