11th Speech in Noise Workshop, 10-11 January 2019, Ghent, BE

Effects of hearing-aid amplification on consonant audibility and forward masking

Borys Kowalewski(a), Johannes Zaar, Michal Fereczkowski, Ewen MacDonald
Hearing Systems, Dept. Electrical Engineering, Technical University of Denmark, Denmark

Olaf Strelcyk
Sonova US Corporate Services, United States

Tobias May, Torsten Dau
Hearing Systems, Dept. Electrical Engineering, Technical University of Denmark, Denmark

(a) Presenting

Dynamic range compression (DRC) is a widely-used hearing-aid compensation strategy. The speed of gain reduction and recovery in a compressor are dictated, respectively, by its attack and release time constants. It has been hypothesized that fast-acting compression, characterized by release times shorter than 200 ms, can provide superior speech audibility and improve the rate of recovery from forward masking in hearing-impaired (HI) listeners. On the other hand, it has been reported that fast-acting compression can lead to distortions of the temporal envelope of the stimuli and degrade speech recognition.

Here, the effects of DRC on HI listeners’ consonant identification in quiet and in interrupted noise were investigated. Several input levels of speech and two compression conditions were considered that differed only in terms of the release time: fast-acting (10 ms release time) and slow-acting (500 ms release time).
A benefit of fast-acting compression was observed at the lowest speech input level in quiet and at medium speech levels in noise. No detrimental effects of fast-acting compression on recognition were found at any of the tested speech levels.

Additionally, the two compensation strategies were evaluated in terms of objective measures such as the output gain, envelope distortion index (EDI) and a metric of consonant audibility. The average amount of temporal envelope distortion was found to be minimal, consistent with the results of the perceptual evaluation. Consonant audibility was found to account for a large part of the variance in the individual performance scores. However, the listeners seemed to differ in how efficiently they use the audible information to correctly identify the consonants.
The results provide more evidence for beneficial effects of fast-acting DRC, at least in a limited class of acoustic scenarios.

Last modified 2018-12-08 00:23:30