Effects of spectral smearing on the identification of speech in noise filtered into low- and mid-frequency regions

Abstract

Léger et al. [J. Acoust. Soc. Am. 131, 1502-1514 (2012)] reported deficits in the identification of consonants in noise by hearing-impaired listeners using stimuli filtered into low- or mid-frequency regions in which audiometric thresholds were normal or near-normal. The deficits could not be fully explained in terms of reduced audibility or temporal-envelope processing. However, previous studies indicate that the listeners may have had reduced frequency selectivity, with auditory filters broadened by a factor of about 1.3, despite having normal or near-normal audiometric thresholds in the tested regions. The present study aimed to determine whether the speech-perception deficits could be explained by such a small reduction of frequency selectivity. Consonant identification was measured for normal-hearing listeners in quiet and in unmodulated and modulated noises using the same method as Léger et al. The signal-to-noise ratio was set to -3 dB for the masked conditions. Various amounts of reduced frequency selectivity were simulated using a spectral-smearing algorithm. Performance was reduced only for spectral-smearing factors greater than 1.7. For all conditions, identification scores for hearing-impaired listeners could not be explained by a mild reduction of frequency selectivity.

Abnormal speech processing in frequency regions where absolute thresholds are normal for listeners with high-frequency hearing loss

“…This may indicate that audibility influenced speech intelligibility, even though the speech was amplified for listeners with PTAs above 20 dB HL in the tested region. This finding is at odd with the results of Léger et al (2012a), for which the audiometric thresholds in the tested regions were not correlated with the speech scores. The values of the extended speech intelligibility index (ESII; Rhebergen and Versfeld, 2005;Rhebergen et al, 2006) were calculated for each listener tested here.…”

“…Simulated mild broadening of the auditory filters had very little effect on speech intelligibility. Léger et al (2012a) concluded that, consistent with previous work of Strelcyk and Dau (2009), slightly reduced frequency selectivity could not entirely explain the speech perception deficits for HI listeners found by Léger et al (2012b).…”

“…The SNR was À3 dB. This SNR was chosen based on related research (Léger et al, 2012a) investigating the effect of the mild reduction in frequency selectivity typically associated with mild forms of hearing loss (e.g., Strelcyk and Dau, 2009).…”

“…However, auditory filters are expected to be slightly broader than normal for HI listeners with audiometric thresholds between 20 and 30 dB HL (e.g., Badri et al, 2011;Moore, 2007;Santurette and Dau, 2007;Strelcyk and Dau, 2009). In a follow-up study, Léger et al (2012a) attempted to determine if the speech deficits could be due to reduced frequency selectivity by simulating a range of broadening of the auditory filters for NH listeners using the spectral-smearing algorithm developed by Moore (1993, 1994). Simulated mild broadening of the auditory filters had very little effect on speech intelligibility.…”

Effects of age and hearing loss on the intelligibility of interrupted speech

“…Previous research suggests strong links between a number of basic auditory abilities in older and hearingimpaired listeners and the perception of speech interrupted by silence or modulated noise. These include forward masking, TFS processing, low-to-mid frequency audibility, and auditory filter bandwidth Nelson, 2006, 2010;Leger et al, 2012;Sheft et al, 2012a). The age-related deficits in temporal processing, however, appeared most detrimental at high interruption rates, while having less of an effect at slower rates F€ ullgrabe et al, 2006).…”