A Castilian Spanish digit triplet identification test for assessing speech intelligibility in quiet and in noise*

Resumo

In Castilian Spanish, unlike in other languages, there exist very few validated tests for assessing speech perception. Here, we present a digit triplet identification test in Castilian Spanish and provide reference results for listeners with normal hearing. The test consists of identifying 100 digit triplets delivered through insert earphones. A triplet is considered to be correctly perceived when each digit is correctly identified and reproduced in the correct position in the triplet. The test includes digits from 0 to 9 uttered by three male and one female speaker. 25% of the triplets were uttered by each of the speakers. Reference psychometric functions (representing percent correct triplet identification against speech level) were obtained for ten young normal-hearing listeners in quiet and for speech-to-noise ratios (SNRs) of 10, 0 and –10 dB. Speech level was varied between 3 and 54 decibels above the listener’s individual mean tonal threshold level at 500 Hz, 1 and 2 kHz (dB SL). The noise was a 32-talker babble. Speech reception thresholds were 5.8, 6.1, 7.2, and 32.2 dB SL in quiet and for 10, 0, and –10 dB SNR, respectively. Ways are discussed to optimize testing time by adjusting the number of digit triplets per condition. The test may be administered automatically using AudioSpeech, a purpose-specific, custom-made application developed in MatlabTM. This software and the digit recordings are available from the authors upon request