Contributing spectral regions to subjective intelligibility of dysphonic speech in noise

Abstract

Dysphonia negatively affects speakers' intelligibility in noisy places. Although this problem is well-recognized, little is known about the relationship between acoustics and intelligibility of dysphonic speech. The purpose of this study was to identify spectral regions critical to the intelligibility deficit in dysphonic speech. Sentences from the Consensus of Auditory-Perceptual Evaluation of Voice were recorded from 18 speakers with dysphonia and 3 age-gender matched speakers with healthy voice. The intensity of the speech samples was normalized at 60 dB SPL, and cafeteria noise was added to these recordings at a signal-to-noise ratio (SNR)-3, 0, and + 3. Perceived (i.e., subjective) intelligibility of these stimuli was rated on a 7-point scale by 45 native speakers of American English with normal hearing. Spearman rank correlation tests were conducted to evaluate the association between subjective intelligibility ratings and spectral energy in the following frequency regions: 0–8 kHz, 8–16 kHz, 0–1 kHz, 1–2 kHz, 2–4 kHz, 4–6 kHz, 6–8 kHz, 8–10 kHz, 10–12 kHz, 12–14 kHz, and 14–16 kHz. The results indicated that the energies in 8–16 kHz, 0–1 kHz, 1–2 kHz, 6–8kHz, and 10–12 kHz ranges were significantly associated with the subjective intelligibility ratings. The clinical relevance of the findings will be discussed.