Zhou Ning, Xu Li
School of Hearing, Speech and Language Sciences, Ohio University, Athens, OH 45701, USA.
Hear Res. 2008 Dec;246(1-2):36-43. doi: 10.1016/j.heares.2008.09.006. Epub 2008 Sep 25.
It has been shown that frequency-place mismatch has detrimental effects on English speech recognition. The present study investigated the effects of mismatched spectral distribution of envelopes on Mandarin Chinese tone recognition using a noise-excited vocoder. In Experiment 1, speech samples were processed to simulate a cochlear implant with various insertion depths. The carrier bands were shifted basally relative to the analysis bands by 1-7 mm in the cochlea. Nine normal-hearing Mandarin Chinese listeners participated in this experiment. Basal shift of the carriers only slightly affected tone recognition. The resistance of tone recognition to spectral shift can be attributed to the overall amplitude contour cues that are independent from spectral manipulations. Experiment 2 examined the effects of frequency compression, where widened analysis bands by 2, 6, and 10 mm were compressively allocated to narrower carrier bands. Five of the 9 subjects participated in Experiment 2. It appears that the expanded frequency information especially on the low frequency end can compensate for the distortion from frequency compression. Thus, spectral shift might not pose a severe problem for tone recognition, and allocation of wider frequency range to include more low frequency information might be beneficial for tone recognition.
研究表明,频率-位置不匹配对英语语音识别有不利影响。本研究使用噪声激励声码器,研究了包络频谱分布不匹配对汉语声调识别的影响。在实验1中,对语音样本进行处理,以模拟具有不同插入深度的人工耳蜗。在耳蜗中,载波频段相对于分析频段向基底方向偏移1-7毫米。9名听力正常的汉语普通话听众参与了该实验。载波的基底偏移对声调识别仅有轻微影响。声调识别对频谱偏移的抗性可归因于独立于频谱操作的整体幅度轮廓线索。实验2研究了频率压缩的影响,即将加宽2毫米、6毫米和10毫米的分析频段压缩分配到更窄的载波频段。9名受试者中有5名参与了实验2。看来,扩展的频率信息,尤其是低频端的信息,可以补偿频率压缩带来的失真。因此,频谱偏移可能不会给声调识别带来严重问题,分配更宽的频率范围以包含更多低频信息可能有利于声调识别。