Conference_programme: 10.3 - Effects of noise on work-related communication disorders

Lecture: Contribution of High-Frequency Energy to Intelligibility of Dysphonic Speech in Noise

Author(s): Ishikawa Keiko , Monson Brian, Khosla Sid, Boyce Suzanne

Summary:
The aim of this project was to examine the relationship between high frequency energy (HFE) and intelligibility deficits in dysphonic speech. Intelligibility can be significantly reduced when conversations take place in the presence of environmental noise. Maintaining acceptable intelligibility in this situation is particularly challenging for speakers with dysphonia. Recent evidence indicates that cues for speech perception reside not only in the frequency range below 8 kHz but also in the higher frequencies. These cues may be especially important for understanding speech when environmental noise masks cues in the lower frequencies. Some dysphonic speech presents with reduced HFE, making the high frequency cues less available to listeners. Accordingly, it was hypothesized that reduced HFE would contribute to intelligibility deficit in dysphonic speech. \n \nSentences from Consensus of Auditory-Perceptual Evaluation of Voice were recorded from 3 speakers with normal voice (1 adult female, 1 adult male1, and 1 child) and 18 speakers with dysphonia (6 adult females, 6 adult males, and 6 children). Cafeteria noise was added after standardizing the intensity of these recordings at three noise levels: signal-to-noise ratio (SNR)-3, SNR+0, and SNR+3. The normal samples were paired with the dysphonic samples in their age and/or gender group, serving as an anchor for the intelligibility rating. These stimuli were presented to 45 native speakers of American English with normal hearing (15 listeners for each noise condition). Intelligibility was rated with a 7-point rating scale. Acoustic analysis was conducted on the stimuli at all noise levels to measure the low and high-frequency energy ratio (L/H), defined as the ratio between the total dB value of summed lower octave bands (62.5 Hz to 4 kHz octaves) and high octave bands (8kHz and 16kHz octaves). Preliminary results indicate moderate negative correlation between the L/H and intelligibility ratings at SNR+0 and +3.

Corresponding author

Name: Dr Keiko Ishikawa

e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Country: United States