10Hearing in Noise and in Space

FWhy quiet scores flatter the implant

Most postlingual adults now reach high open-set sentence scores in a quiet booth, so quiet testing alone can no longer separate good from struggling users; it has hit a ceiling for the average recipient. Performance falls steeply once background noise is added, and the rate of decline with worsening signal-to-noise ratio is far greater for implant users than for normal-hearing listeners. Datalogging shows much real-world speech arrives below 60 dB SPL, softer and more distant than the 65 to 70 dB SPL levels used in standard test protocols, widening the lab-to-life gap. Fluctuating babble of the kind found in restaurants, classrooms and parties is harder than steady noise, because implant users cannot listen in the dips the way normal-hearing listeners do. Difficult listening also costs effort and fatigue that a percent-correct score never captures, which is one reason ecological-validity testing is now emphasised.[2025][2020][2009]

TMeasuring speech in noise: the adaptive SRT

The speech-reception threshold in noise is the signal-to-noise ratio at which a listener correctly identifies 50 percent of the speech material; it is reported in dB SNR, and a lower (more negative) value is better. Adaptive procedures vary the SNR trial-by-trial, raising it after errors and lowering it after correct responses, to home in on the 50-percent point efficiently rather than testing at one fixed SNR. Normal-hearing listeners reach thresholds near 0 dB or several dB negative for sentences in noise; typical implant users need a much more favourable, positive SNR to reach the same 50-percent point. Fixed-SNR percent-correct tests (often at +10 or +15 dB) and adaptive threshold tests probe the same ability but report it differently; the adaptive threshold avoids ceiling and floor effects across a wide performance range. Directional microphones and noise-reduction algorithms are designed to improve the input SNR before coding, and give their largest measurable benefit in steady, spatially separated noise.[2025][2020][2017]

CSpatial release from masking

Spatial release from masking is the improvement in the speech-reception threshold when the masker is moved away from the target in space rather than sharing the target's location; it is the benefit of spatial separation expressed in dB. Normal-hearing listeners gain a large spatial release, often around 10 dB or more, by exploiting head-shadow level differences plus binaural processing of interaural time and level cues. A single implant captures mainly the monaural head-shadow component when the noise sits on the far side; it gains little when target and masker are both in front, because one ear cannot exploit interaural cues. Spatial release therefore shrinks markedly with one implant, and shrinks further when the masker is on the same side as the device, which is why unilateral users seek to seat themselves with noise on the implanted side.[2017][2020]

CLocating sound with one ear

Horizontal localisation depends on comparing the two ears: interaural time differences dominate for low frequencies and interaural level differences for high frequencies, so a single ear has almost no usable horizontal cue. Localisation accuracy is reported as the root-mean-square error in degrees between the perceived and actual source angle; smaller is better, and normal-hearing performance is on the order of a few degrees. Unilateral implant users show large localisation errors, frequently tens of degrees and often near chance for left-right judgements, relying on weak monaural spectral and loudness cues rather than true binaural comparison. Front-back and same-side discriminations are the hardest, and the absence of localisation has real safety and social cost even when speech scores look good. This monaural limit is the central argument for a second ear, addressed by bilateral, bimodal and single-sided-deafness fittings.[2020][2009]