Relative importance of different spectral bands to consonant identification: Relevance for frequency transposition in hearing aids

Abstract

Listeners with high-frequency dead regions (DRs) benefit from amplification of frequencies up to 1.7 times the edge frequency, f(e), of the DR. Better consonant identification might be achieved by replacing the band from f(e) to 1.7f(e) with a higher spectral band. We aimed to identify the optimal band, using simulations with normal-hearing listeners. In experiment 1, nonsense syllables were lowpass filtered to simulate DRs with f(e) of 0.5, 0.75, and 1.0 kHz. Identification was measured for each of these base bands alone and with a bandpass-filtered band added (but not transposed). The added band either extended from f(e) to 1.7f(e) or its center frequency was increased, keeping bandwidth fixed in ERB(N)-number. Performance improved with increasing center frequency and then reached an asymptote or declined. Experiment 2 used a mid-frequency base band, and a lower-frequency added band. The results also showed a beneficial effect of frequency separation of the added and base bands. Experiment 3 resembled experiment 1, but with bandwidth fixed in Hertz. For higher-frequency added bands, the benefit was lower than for experiment 1.

The importance for speech intelligibility of random fluctuations in “steady” background noise

“…Also, the ability to process the envelope fluctuations of the target may have been adversely affected by the larger envelope fluctuations of the background (given the -7 dB TBR), via the phenomenon of modulation detection/discrimination interference (Yost et al, 1989;Moore et al, 1991;Buus and Pan, 1994). Interestingly, the adverse effect of amplitude fluctuations in the background for EDM occurred despite the fact that the fluctuations of the background were the same across channels for the SAM background, and would have been partially correlated across channels for the Tlk background (Crouzet and Ainsworth, 2001;Vickers et al, 2009). One might have expected that the coherence of envelope fluctuations across channels would promote the perceptual fusion of the envelope signals for the backgrounds and hence improved the segregation of the target and background (Bregman et al, 1985;Hall and Grose, 1990).…”

Predicted effects of sensorineural hearing loss on across-fiber envelope coding in the auditory nerve

“…Acoustical analyses indicate that for very low-rate modulations (<5 Hz), the correlation of envelope information between most spectral channels is very high (>0.8) (Crouzet and Ainsworth, 2001; see also Houtgast, 1999 andVickers et al, 2009). This high correlation reflects the syllabic rate in speech (i.e., syllable onsets and offsets are similar across the speech spectrum) (Arai and Greenberg, 1997).…”

“…This approach provides a quantitative physiological framework that may be beneficial for expanding current methods for predicting speech intelligibility (e.g., the speech transmission index, STI, Steeneken and Houtgast, 1980;Houtgast and Steeneken, 1985) by considering physiological envelope responses (Elhilali et al, 2003), including the effects of cross-channel correlation (Steeneken and Houtgast, 1999;Vickers et al, 2009).…”

Effect of linear and warped spectral transposition on consonant identification by normal-hearing listeners with a simulated dead region

“…The experiments of Vickers et al (2009) appear to suggest that transposition of a source band to a destination band with the same width in ERB N -number might be more benefi cial than transposition of a source band to a destination band with the same width in Hertz. However, their experiments did not actually involve transposition.…”

“…Therefore, the frequency components in the source band, when transposed to the destination band, might provide more information than the original components in the destination band, leading to greater intelligibility. Vickers et al (2009) evaluated the potential benefi t of such an approach using normal-hearing listeners. One aim of their work was to identify the centre frequency of the source band that provided the most information (when it was not transposed).…”