Temporal and spectral masking release in low- and mid-frequency regions for normal-hearing and hearing-impaired listeners

Abstract

"Masking release" (MR), the improvement of speech intelligibility in modulated compared with unmodulated maskers, is typically smaller than normal for hearing-impaired listeners. The extent to which this is due to reduced audibility or to suprathreshold processing deficits is unclear. Here, the effects of audibility were controlled by using stimuli restricted to the low- (≤1.5 kHz) or mid-frequency (1-3 kHz) region for normal-hearing listeners and hearing-impaired listeners with near-normal hearing in the tested region. Previous work suggests that the latter may have suprathreshold deficits. Both spectral and temporal MR were measured. Consonant identification was measured in quiet and in the presence of unmodulated, amplitude-modulated, and spectrally modulated noise at three signal-to-noise ratios (the same ratios for the two groups). For both frequency regions, consonant identification was poorer for the hearing-impaired than for the normal-hearing listeners in all conditions. The results suggest the presence of suprathreshold deficits for the hearing-impaired listeners, despite near-normal audiometric thresholds over the tested frequency regions. However, spectral MR and temporal MR were similar for the two groups. Thus, the suprathreshold deficits for the hearing-impaired group did not lead to reduced MR.

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

“…However, as found in previous work Lorenzi et al, 2009;Strelcyk and Dau, 2009), the deficits were not large (10e25% correct or 2e3 dB in speech reception threshold, SRT), and their impact in daily listening situations is unclear. Nevertheless, Léger et al (2012b) noted that four out of the nine HI listeners initially tested using the midfrequency region were excluded from the main experiment because their speech scores in quiet were too low (below 35% correct) to allow meaningful performance for speech in noise. This suggests that Léger et al (2012b) may have underestimated the typical speech identification deficits shown by HI listeners with hearing loss above 3 kHz when listening to stimuli filtered into their regions of normal hearing (<3 kHz).…”

“…Nevertheless, Léger et al (2012b) noted that four out of the nine HI listeners initially tested using the midfrequency region were excluded from the main experiment because their speech scores in quiet were too low (below 35% correct) to allow meaningful performance for speech in noise. This suggests that Léger et al (2012b) may have underestimated the typical speech identification deficits shown by HI listeners with hearing loss above 3 kHz when listening to stimuli filtered into their regions of normal hearing (<3 kHz). The large variability across listeners may indicate that a variety of phenotypes underlie high-frequency hearing loss, consistent with recent work investigating individual differences among HI listeners (for a review of different subtypes of prebycusis, see Schmiedt, 2010).…”

“…Following Lorenzi et al (2009), Léger et al (2012b argued that the general speech perception deficit observed for the HI listeners tested in frequency regions in which audiometric thresholds were normal or near-normal could not be due to difficulties in processing the temporal envelopes of the stimuli. Indeed, temporal masking release (the difference in identification for speech in a notionally steady and an amplitude-modulated noise) was similar for the NH and HI listeners.…”

“…Over the last decade, several studies Léger et al, 2012b;Lorenzi et al, 2009;Strelcyk and Dau, 2009) have investigated speech processing for hearing-impaired (HI) listeners using stimuli filtered into a low-frequency region (<1e2 kHz) where audiometric thresholds were normal or near-normal according to audiological standards (i.e., better than 25e30 dB HL; Goodman, 1965). This approach has been used to demonstrate and characterize the specific contribution of suprathreshold auditory deficits (e.g., reduced frequency selectivity and reduced temporal resolution) to speech intelligibility for HI listeners, while limiting the influence of reduced audibility.…”

“…This approach has been used to demonstrate and characterize the specific contribution of suprathreshold auditory deficits (e.g., reduced frequency selectivity and reduced temporal resolution) to speech intelligibility for HI listeners, while limiting the influence of reduced audibility. Léger et al (2012b) filtered stimuli (speech þ masking sound) to restrict their spectra to regions for which the HI listeners showed normal ( 20 dB HL) or near-normal ( 30 dB HL) audiometric thresholds. Two regions were tested: a low-frequency region ( 1.5 kHz) and a mid-frequency region (1e3 kHz), and both normal hearing (NH) and HI listeners were tested for each frequency region.…”