Giguère Christian, Berger Elliott H
a Audiology and Speech-Language Pathology Program, University of Ottawa , Ontario , Canada and.
b 3M Personal Safety Division, E•A•RCAL Laboratory , Indianapolis , USA.
Int J Audiol. 2016;55 Suppl 1:S30-40. doi: 10.3109/14992027.2015.1129460.
To investigate the effects of hearing protection on speech recognition in noise.
Computational study using a speech recognition model that was previously empirically validated.
Recognition scores were calculated in unprotected and protected conditions for four sets of hearing protector attenuation functions in two different noises, for three simulated hearing profiles illustrative of those anticipated in the noisy workplace.
For a normal-hearing profile, recognition scores were not sensitive to the slope of the attenuation function and the overall amount of noise reduction, but protected conditions provided a small but consistent 7-12% benefit compared to unprotected listening. For profiles simulating hearing loss, recognition scores were much more sensitive to the attenuation function. Substantial drops of 30% or more were found compared to unprotected listening in some conditions of steep attenuation slopes and large noise reductions. Attenuation functions modelled from real hearing protectors with nearly-flat attenuation yielded a benefit compared to unprotected listening for all hearing profiles studied. These findings were true in both noises.
Limiting the slope of the hearing protector attenuation function and/or the overall amount of noise reduction is useful and warranted for workers with hearing loss to prevent adverse effects on speech recognition.
研究听力保护对噪声环境下语音识别的影响。
使用先前经实证验证的语音识别模型进行计算研究。
针对两种不同噪声中的四组听力保护器衰减函数，在未保护和保护条件下，为三种模拟听力状况计算识别分数，这些听力状况代表了嘈杂工作场所预期的情况。
对于正常听力状况，识别分数对衰减函数的斜率和总体降噪量不敏感，但与未保护聆听相比，保护条件下有7%至12%的小幅但持续的优势。对于模拟听力损失的状况，识别分数对衰减函数更为敏感。在一些陡峭衰减斜率和大量降噪的条件下，与未保护聆听相比，发现识别分数大幅下降30%或更多。对于所有研究的听力状况，与真实听力保护器建模的几乎平坦衰减的衰减函数相比有优势。这些发现在两种噪声中均成立。
限制听力保护器衰减函数的斜率和/或总体降噪量对听力损失的工人是有用的且有必要的，以防止对语音识别产生不利影响。
