Young L L, Goodman J T, Carhart R
J Am Audiol Soc. 1978 Jan-Feb;3(4):167-71.
Consonant-nucleus-consonant monosyllabic words were filltered such that each spectral component had equal energy (i.e., "whitened") and peak clipped in one of four ways: minimal, 20, 30, and 40 dB of clipping. In addition, unmodified consonant-nucleus-consonant words were used as stimuli. These different types of sppech were presented to 20 persons with normal hearing at various sensation levels. The results indicate that whitening and peak clipping do not substantially degrade speech intelligibility. In fact, under some conditions whitening and peak clipping may slightly enhance intelligibility.
辅音-元音-辅音单音节词经过滤波处理,使每个频谱成分具有相等的能量(即“白化”),并以四种方式之一进行峰值限幅:最小限幅、20 dB限幅、30 dB限幅和40 dB限幅。此外,未修改的辅音-元音-辅音词被用作刺激音。这些不同类型的语音在不同的感觉水平下呈现给20名听力正常的人。结果表明,白化和峰值限幅不会显著降低语音清晰度。事实上,在某些情况下,白化和峰值限幅可能会略微提高清晰度。