Speech intelligibility with symmetrically-placed interferers for German- and Mandarin- speaking listeners in anechoic and reverberant conditions
Although the semantic information is also expressed by pitch contour in tonal languages, it is not clear whether tonal language cochlear implant (CI) listeners perform worse than western CI listeners if the CI coding strategy only delivers envelope information. To investigate possible language-specific effects in speech intelligibility (SI) and spatial release from masking (SRM), referring to the effect that listeners benefit when the target speaker is spatially separated from interfering sources in comparison to co-located target and interferers, speech reception thresholds (SRTs) were obtained. Co-located and symmetrically placed maskers (±60°) were used in connection with the German matrix test (OLSA) and the newly developed Mandarin Chinese matrix test (CMNmatrix, Hu et al., 2018, IJA), for both vocoded and non-vocoded signals in two groups: native German-speaking (GS) listeners and Mandarin-Chinese-speaking (CMNS) listeners. SRTs were tested using headphone presentation in a sound attenuated booth with either a stationary masker or a fluctuating nonsense speech masker. The co-located and separated conditions in three different rooms (anechoic, 0.6 s, and 3 s reverberation time) were simulated by using virtual acoustics and headphone auralization. For the anechoic room, an artificial infinite interaural level difference (ILDinf) condition, where the acoustic crosstalk was removed, was additionally tested.
For the anechoic room, the results were comparable to those reported in (Hu et al., JASA, 2018): both groups greatly benefited from spatial separation and ILDinf for the non-vocoded signals, while no binaural benefit was observed in noise vocoder simulated BiCI listeners for the ±60° spatial configuration. CMNS listeners showed a slightly reduced binaural benefit in both ±60° and ILDinf conditions relative to the co-located condition when compared to GS listeners. As expected, speech intelligibility decreased in reverberant conditions compared to the anechoic condition. Increasing reverberation time showed less effect on the SRM in CMNS listeners than in GS listeners. The potential role of pitch contour perception and the role for CI processing are discussed.