Use of binaural cues for speech understanding with different masker types
Benjamin Dieudonné, Tom Francart · Lirias (KU Leuven) · 2016
Introduction. Spatial separation of speech and noise improves speech intelligibility, due to interaural time and level differences (ITDs and ILDs respectively). This improvement is less for hearing- impaired listeners; therefore, it is important to know how and which cues should be enhanced to alleviate this deficit. Spatial separation yields one ear with a better signal-to-noise ratio (SNR) than the other: the “better ear”. In addition, listening with two ears (instead of the better ear only) yields an extra binaural benefit in speech intelligibility: squelch. Mostly, squelch is attributed to binaural decorrelation. This is possible due to ITDs yielding different correlation peaks (separated in time) for target and masker. Moreover, central processing may be involved as well: the ability to better segregate target and masker based on the perception of their spatial location. For the latter, both ITDs and ILDs provide the ability to localize sounds. However, when only ILDs are present, squelch is not observed when using speech-weighted noise (SWN) as a masker [1]. We investigated whether squelch may be present with only ILDs when using a fluctuating masker. Hypotheses. Firstly, we hypothesize that ILD-only squelch may be observed if the masker is modulated noise. During temporal gaps in the masker, the better and worse ear are equivalent in SNR; therefore the listener may benefit from binaural redundancy. Secondly, we hypothesize that ILD-only squelch may also be observed if the masker is a competing talker. We hypothesize that the spatial location itself (as opposed to the physical cues induced by spatial location) may contribute to spatial release from masking if the masker is not merely energetic. Methods. A Dutch sentence test was used to measure speech intelligibility. We investigated three different maskers, spectrally matched to target speech: SWN, a competing talker and modulated SWN with the same temporal envelope as the competing talker. Spatial hearing was simulated with head-related transfer functions; ITDs were set to zero for the ILD-only conditions (and vice versa). Results. In preliminary results we did not observe ILD-only squelch with the modulated SWN. We did however observe ILD-only squelch with a competing talker. Conclusion. Squelch may be observed with only ILDs if the masker is a competing talker. Therefore, it may be advisable to preserve ILDs in binaural hearing aids for better speech intelligibility.